Download economic growth in paraguay - Global Development Network

ECONOMIC GROWTH IN PARAGUAY*
Carlos G. Fernández Valdovinos
Alexander Monge Naranjo
Gerencia de Estudios Económicos,
Department of Economics,
Banco Central de Paraguay
Northwestern University
May , 2003
Abstract.
This paper explores the economic growth in Paraguay from 1962 to the
present. We discuss in detail the different episodes as policies and growth results
were very different across periods. Overall, the weakest accomplishments of the
Paraguayan economy are the accumulation of human capital and, more astonishingly,
in the behavior of total factor productivity (TFP). In fact, using a wide variety of
methodologies, we argue that TFP continuously declined after mid 1970s.
The paper also explores the relationship of growth with social indicators,
especially poverty, and argues for more aggressive government policies in promoting
the accumulation of human capital. Those policies would be likely to improve the
country not only in terms of poverty but also in its growth performance.
To explore the effect of the (dramatic) fluctuations on the long-term growth
of Paraguay, we include an analysis of the business cycles in Paraguay during the
different periods, including its co-movements with its main trading partners. We
observe that the importance of Brazilian fluctuations has increased dramatically in
the recent years.
*
We are very thankful to Rodolfo Manuelli and Victor Elias for their detailed comments and to Bernardo Darío Rojas Páez for
research assistance. The authors are solely responsible for the contents of this paper. Specially, they do not necessarily represent the
opinions of the Banco Central de Paraguay.
The series used in this paper and the methodologies used in their construction are available upon request to the authors.
1. Introduction.
This paper studies the growth behavior of Paraguay, a small economy in South America that is among the
least developed in the American continent. According to the index of human capital development,
Paraguay lags significantly behind most of its neighbors. The current conditions of Paraguay, which are
the likely outcome of a long history of unsustainable macroeconomic imbalances, policy reversals, bad
policies and political repression and turmoil, can be ascertained by looking at the current statistics on
poverty. The 1999’s Household Survey 1 reveals that poverty in the urban regions of the country reached
26.7%. This means that around 810,000 individuals in these areas can barely finance their daily food
expenditures. Moreover, around 6.1% of the urban population are into the “extreme poverty” class, this
is, 184,000 people cannot cover expenditures for basic food. The situation is even more dramatic in the
rural sector: 42% of total families fell below the poverty line, of which 26.5% were below the extreme
poverty line. Even by Latin American standards, the development conditions in Paraguay are dire.
Isolated by nature, with a small and unskilled population, lacking mineral resources, punished by
devastating wars and continuously disturbed by the macro unstability of its neighbors, Paraguay is a
natural candidate to be among the least developed countries in Latin America. As of 2000, its GDP per
capita was only 50% of the Latin American average and, when compared with its MERCOSUR partners,
it was only 34%. Worse of all, the Paraguayan economic growth performance has been historically less
than satisfactory. The average annual growth rate between 1950 and 2000 was only 1.7%. If we exclude
the seventies, when much of the construction of the Itaipú project was done, the average drops to only
0.5% per year2. The “lost decade” of the eighties, which was suffered by all the continent, left Paraguay
with a yearly GDP per capita growth rate of only 0.2%. But contrary to the average Latin American
country, in Paraguay that period was followed by an even worse decade: in the nineties the average
annual growth rate in GDP per capita was negative, –1.7%. Current per capita GDP levels are equal to
those in 1976.
Without any doubt, the meager growth rates displayed by the Paraguayan economy lie at the center of all
other major problems. To put things in perspective, for example, between 1990 and 1995, per capita
Chilean GDP grew in average 5.3% per year while in Paraguay the average annual growth rate was only
0.5%. Moreover, expressed in 1990 dollars, average GDP per capita in Latin America was USD 2604 in
1995 and only USD 1466 in Paraguay3. Thus, starting from its 1995 level, it will take a little more than
11 years for Paraguay to reach the Latin American average GDP if it could grow at the Chilean rates.
1
The survey is elaborated by the Dirección General de Estadísticas Encuestas y Censos (DGEEC).
During the seventies, per capita GDP grew in average at a record rate of 6.2% per year. Growth rate in per capita GDP reached even
8.6% in 1978.
2
Instead, with its own growth rates, it would take 192 years to attain the same goal. This example
illustrates an obvious point: growth rates have cumulative effects over time and even small differences in
the growth rates, when cumulated over a generation or more, have much huge consequences for standards
of living. If anything, macroeconomists and other policy makers in the region must have growth policies
as their first priority.
In this paper we examine the Paraguayan growth experience from early 1960s to these days. We study
the different episodes, attempting to identify the main factors behind the growth exhibited by Paraguay in
each episode. Studying the determinants of growth is of particular importance for a country like Paraguay
that just recently recovered the most basic political rights and, with some vacillation, tries to consolidate
democratic institutions. Up to now, those positive political events have not been accompanied by
economic growth and poverty alleviation.
Over the different periods, Paraguay shows very different growth rates. The following table compares the
growth performance of Paraguay between 1960 and 1995 relatively to MERCOSUR countries, Latin
America countries and the world. It can be seen that the average growth performance of Paraguay’s has
not been impressive. Even though higher than the Latin America and MERCOSUR average, it was way
below the mean of East Asia and OECD countries. Moreover, growth in Paraguay has been very erratic
over time. The sixties were a middling decade, with Paraguay growing in tandem with the region. During
the seventies, the construction of Itaipú, a huge hydroelectric project, allowed Paraguay to growth at the
same high rates observed in East Asia, clearly outperforming the rest of Latin America. For the eighties
this stimuli vanished and, as in other countries in the region, average per capita growth rate was negative.
As it is well known, this period was characterized by external debt problems, stagnation and
macroeconomic instability in the whole region. But in the nineties, Paraguay took apart from the
continent. While Latin America observed a strong recovery, Paraguay experienced probably the worst
growth episode in recent history: its average growth rate during this period was around 1/3 of the Latin
America and 1/5 of its MERCOSUR partners.4
3
These numbers are taken from the Inter American Development Bank.
Following a banking crisis and an strong reduction in triangular trade, economic performance was even worst in the second part of
the decade. Between 1995 and 1999, the economy contracted for four consecutive years, with an overall fall in real per capita GDP of
6.5 percent, one of the worst macroeconomic performance in all Latin America.
4
TABLE 1:
GDP Per Capita and Average GDP Per Capita Growth Rates
1960
1995
60-70
70-80
80-90
90-95
60-95
Paraguay
1,177
2,178
1.7
6.2
-1.7
0.5
1.8
MERCOSUR*
2,571
4,132
2.3
3.0
-1.2
2.5
1.5
Latin America (21)
2,319
3,429
2.3
2.3
-1.5
1.4
1.1
784
1,061
2.1
1.1
-0.8
-1.9
0.5
East Asia
1,275
8,119
4.7
6.0
4.6
4.1
5.0
OECD (22)
5,592
13,364
4.3
2.5
2.1
1.1
2.7
World (81)
2,667
6,141
3.2
2.6
0.6
1.1
2.0
Sub-Saharan Africa (17)
* Includes Bolivia and Chile
Source: De Gregorio and Lee (1999).
The most striking conclusion from the growth accounting exercises is that total factor productivity has been
falling over time. This result is robust to a variety of methods to measure capital and to decompose input
contribution versus productivity. We also find that physical capital had a strong pull for growth in most of the
periods, but it was more strongly during the 1970s and much more modest during the 1990s. We believe that
those are the reflections of the construction of the large Itaipu dam in the 1970s and of the effect of the
political uncertainties after 1989 when Stroessner was ousted out of power.
Over the entire sample period, it is clear that capital accumulation has outpaced output, so its accumulation
does not seem to be the major deterrent of growth. The main problems appear in the accumulation of human
capital and the overall productivity of factors. It seems safe to conclude that if Paraguay is to grow faster,
more aggressive policies should focus on these problems. Any contribution would also have the benefit of
fostering the accumulation of complementary factors, such as physical capital.
The remainder of the paper proceeds as follows. In the next section we provide an overview of the most
important historical events during the second part of the past century. In the third section we examine the
statistics
on human capital and poverty and argue that Paraguay’s performance in this front is very
disappointing, even if compared with the countries of the region. In the fourth section we importance of
factor accumulation and productivity in leading the growth of output. In the fifth section we argue that fiscal
variables, inflation and some other economic variables have a strong co-movement with growth and hence are
likely to influence it. The sixth section looks at the comovement of Paraguay with its main trading partners
and argues that the importance of Brazil has increased dramatically in recent years. In the seventh section we
include some final remarks. Finally, an appendix is included with a discussion of the major factors affecting
the incentives to accumulate capital.
2. A Brief Historical Background.
Paraguay has always been an economy concentrated in the production of agricultural goods. This
characteristic and the small size and openness of the country, made it very sensitive to world events that
affected the international market for agricultural products. Those events translated into sharp and long
lasting fluctuations that have also triggered other macro instabilities, such as fiscal and exchange crisis
and high inflation rates. The purpose of this section is to briefly review the main characteristics of the
different sub-periods from 1940 to the present.
2.1. The Forties, Fifties and Sixties.
The forties and fifties were periods of important institutional changes and severe fluctuations. The period
started with World War II, which mostly benefited Paraguayan in the form of a sharply increased in the
demand for its agriculture products, leading to relatively high growth rates of the entire economy5.
Indeed, between 1938 and 1946, average GDP grew 2.5% per year and exports grew at an annual rate of
8.2%. But world supply conditions and the higher demand of domestic output were also reflected in
rising domestic prices. Between 1939 and 1944 the cost of living increased 300% for higher income
groups and 50% for the poor. In the institutional front, as in most countries in Latin America, this is the
period when Paraguay established its national currency and founded its Central Bank. Effectively, in
November 1943 the Guaraní became the country’s monetary unit, with an initial exchange rate of G. 3.07
per US dollar.
But what appeared to be the beginning of a new and better era for the country came to a sudden end.
Effectively, the cessation of world hostilities in 1945 brought about a long lasting and drastic drop in
demand for agricultural products. Additionally, a cruel civil war erupted in 1946. The consequences were
dire. For instance, output in 1947 dropped by 13%. The problems caused by low export demand and
social unrest that disrupted production not only continued in the early fifties, but also they were further
exacerbated by imprudent financial policies. In an attempt to encourage production, credit policies
became expansive, fueling inflation and draining foreign reserves. Inflation accelerated reaching 160% in
1952 and more distortions were introduced in the economy as a system of multiple exchange rate and
exchange rate controls gradually developed. With the aid of an IMF mission, the government was later
able to bring inflation under control and to stabilize the currency at an exchange rate pegged to the US
5
During this period, the volume of tobacco exports tripled, vegetable oil increased more than six times, wood multiplied five times,
meat exports doubled while cotton increased 50%.
dollar (G. 126 per dollar)6. Growth recovered in the latter half of the fifties with commerce and
construction as the most dynamic activities.
During the fifties foreign aid was ample and basic physical infrastructure was gradually expanded. In this
respect, it is important to highlight the treaties signed with Brazil. In January 1956, Paraguay signed an
agreement with this country whereby Brazil offered to finance the studies and to consign the loans
necessary for the construction of a hydroelectric plant in the Acaray River, located close to the area
where transport projects were being advanced. This treaty is the first one of its type. The other, which we
will review in detail below, was the construction of Itaipú and it led the country to a large period of
growth.
TABLE 2: Sectoral Composition of GDP (%)
Sectors
1951/60
1961/70
1971/80
1981/90
1991/00
Agriculture
38.1
34.5
29.3
26.2
26.8
Mining
0.1
0.1
0.2
0.4
0.5
Manufactures
16.7
17.1
17.7
16.5
14.7
Construction
1.5
2.2
3.5
6.2
5.4
Electricity, Gas, Water
n.a.
0.6
1.4
2.6
5.2
Transport and Comm.
n.a.
4.2
4.2
4.4
4.9
Commerce and Finance
26.0
25.8
26.3
26.7
24.9
Government
4.3
4.4
4.5
4.4
5.3
Misc. Services
13.3
11.1
12.9
12.6
12.3
Source: Central Bank of Paraguay.
As stated above, agriculture has been the main activity in Paraguay since colonial times. The forties and
fifties were not the exceptions. In 1960 agriculture still accounted for almost 39% of GDP and employed
55% of the economically active population. Manufactures contributed 17.3% to the GDP, employing
15% of the work force. Over 75% of the value-added in manufactures was originated in agro industries.
During the first half of the sixties, agriculture continued to be the main stimulus to economic growth.
Agricultural growth, in turn, reflected migration to the eastern part of the country, including Brazilian
immigration, and the expansion of transport links with Brazil and the internal road network.
The second half of the decade was very different. Public sector works and commerce started leading the
rhythm of growth7. The government started to carry out important programs, especially road construction,
6
Between 1952 and 1955 the average inflation rate was 53% per year. After the implementation of the stabilization plan, the increase
in the price level was reduced to 12.2% per year between 1956 and 1960.
7
For example, during this period construction was increasing at annual rates well above GDP´s, while electricity and water growth
began to exceed GDP´s.
hydropower development, expansion of port facilities, construction of water services for Asuncion and
even the construction of a cement plant. As a consequence, public investment averaged 5.3% of GDP in
1966-70, which is double the rate of the previous five years. The increased investment was partly
financed externally but mostly from higher public savings originated from increases in taxes. Among
public investment, energy ventures to take advantage of the country’s hydroelectric potential were the
most striking. Besides the construction of the Acaray plant, Paraguay and Brazil signed in June 1966 the
Acta Final, the basis for the Itaipú treaty which was finally signed in 1973.
Commerce was another growing sector. It expanded by more than 6% per year during the period. The
main basis for this sector were tourists from Argentina and Brazil. They were being attracted by the lower
costs of goods, driven by lower prices of non-tradables and the much lower taxes that Paraguay imposed
on imported goods, compared to the protectionist policies of richer Argentina and Brazil.
With savings and investment hovering around 12-13% of GDP, the average growth rate of real GDP in
the 1960s was 4.2% per year, while population was growing at 2.5%. After the financial chaos of the
1947-54 period, stability was restored and in the sixties the country experienced price stability with the
cost of living rising at an average of 2% per year. Additionally, during the decade there was a total
exchange rate stability as the country stuck to the same pegged exchange of G. 126 per dollar.
2.2. The Seventies and the Itaipú Phenomenon.
The seventies produced unusual prosperity for Paraguay. During this period, GDP growth accelerated
dramatically to an average of nearly 9% per year, doubling the average performance from the previous
decade8. The driving force behind this growth came from two sectors: agriculture and construction. The
former was the result of the expansion of the agricultural frontier and the latter a consequence of the
surge of construction on various infrastructure projects, culminating in the building of Itaipú (jointly with
Brazil), the world’s largest hydroelectric project.
8
In particular, the yearly real growth rate of the GDP averaged over 11% in the 1977-1980 period.
TABLE 3: Annual Average Growth Rate of Components of GDP (%)
Sectors
1951/60
1961/70
1971/80
1981/90
1991/00
Agriculture
1.8
3.0
6.7
4.0
1.7
Mining
n.a.
57.5.
28.4
4.9
2.7
Industry
1.9
6.5
8.3
2.2
0.8
Construction
7.5
7.4
20.3
0.7
2.8
Electricity, Gas, Water
n.a.
11.3
17.5
7.9
10.2
Transport and Comm.
n.a.
37.3
9.7
3.7
4.1
Commerce and Finance
3.5
4.8
9.0
2.8
-0.5
Government
3.3
7.8
4.3
5.0
5.4
GDP
2.9
4.8
8.8
3.1
2.0
Source: Central Bank of Paraguay.
Starting in early 1970s the government accelerated efforts to expand the agricultural frontier. Heavy
investment in infrastructure began during those years and they opened up the eastern frontier of the
country to development. During the seventies new lands in this fertile region were brought under
cultivation through establishment of numerous settlements there9 10. These settlements were the main
reason for agricultural growth, a sector that reached an average growth rate of 6.9% per year during the
decade. Due to expanding world demand and favorable international prices, the frontier lands were
mainly used to produce export crops, mostly cotton and soybeans, which came to dominate the country’s
exports. As a result, cotton rose from supplying only 1.1% of total exports in 1960 to 44% in 1985, while
soybeans, which did not appear on export lists at all in 1960, attained a share of over 16% in 198111. It
should be noted that livestock and quebracho extracts, Paraguay’s traditional exports, declined
dramatically during the same period.
The highway to Brazil and the development of the lands in the frontier region substantially reduced
Paraguay’s traditional dependence on Argentina as its trade route. This was reflected by the dramatic
increase in trade with Brazil in detriment with the trade with other countries, especially the United States.
For example, in 1960 only 0.2% of Paraguay’s exports went to Brazil and 0.8% of import come from
there. By 1981 the respective percentage were 18.3% and 25.9%. Another impact of the expanding
agricultural frontier was reflected on the regional distribution of the population. While only 18.3% of the
population lived in the eastern frontier region in 1962, 27.3% lived there 20 years later. Also, about
9
By the end of 1976, almost 90,000 land titles had been issued, covering about 4 million ha.
Paraguayan colonists were joined by large numbers of Brazilian and Japanese farmers, who came to the area in response to a
number of economic stimuli: low land prices, low taxes, and favorable international prices for the agricultural products.
11
The importance of soybeans exports, as percentage of total exports, further increase during the eighties and nineties.
10
40.5% of the population lived in the minifundia region in 1962, compared to 34.2% in 198212. It is very
important to note that because of this, and unlike most other LDCs, Paraguay did not experience a
marked urban-rural migration, but instead a rural-rural movement, out of the traditional minifundia
regions to the newly opened lands.
TABLE 4: Employment by Sector of Production (%)
Sectors
1950
1962
1972
1982
1992
1999
Agriculture
55.4
54.7
47.9
42.9
35.6
30.4
Industry
16.1
15.1
14.0
12.0
12.5
12.3
Construction
3.0
3.3
3.9
6.7
7.2
5.2
Transport and Comm.
2.3
2.5
2.8
2.9
3.3
4.3
Commerce and Finance
7.1
7.1
8.0
9.3
13.9
24.8
Services
16.0
17.3
23.4
26.2
27.5
23.0
Source: Population Censuses and Household Surveys. Different years.
The other major source of growth in the seventies was the construction of Itaipú, the largest hydroelectric
dam in the world. The works, which were carry out mainly between 1973 and 1983, cost more than four
times Paraguay’s GDP and were financed externally trough the Itaipú Binational Entity. The debt was
guaranteed by Brazil13. It has been estimated that between 1977 and 1980 around USD 250 million (6%
of its GDP) were spent in Paraguay each year.
The capital inflow originated by the Itaipu project and the easy credit conditions in the international
markets of the time translated in large increases of liquidity and a tremendous credit expansion, while at
the same time it put downward pressures on the real exchange rate. Internally, investment construction
benefited most from the easier credit policies, growing at an average annual rate of 23% between 1973
and 1981. The dramatic increase in construction stemmed from a large expansion in private investment
that represented less than 10% of GDP in the sixties and over 20% at the end of the seventies14. At the
same time, public investment remained at about 5% of GDP.
However, linkages to other sectors of the economy were weak. The exception was the service sector,
especially commercial and financial activities. For example, in 1972 there were only six banks in
Asunción. By 1981 there were 20. As a consequence, commerce and finance grew at annual rates of over
10% in the 1976-80 period and this sector accounted for 26% of GDP by 1981.
12
The minifundia region is defined as the four Departments surrounding Asunción, that is Cordillera, Guairá, Paraguarí and Central.
Most of this spending is not shown in Paraguay’s national accounts since for this purposes Paraguay does not consider binational
enterprises as being located within the national territory.
13
Public finances remained strongly during this period and huge increases occurred in foreign reserves,
which grew from less than USD 20 million at the beginning of the seventies to USD 800 million in 1981.
While Itaipú construction created a substantial increase in effective demand, an increase in the supply of
domestic consumer goods was not immediately forthcoming and the excess demand was only partially
met by increased imports. Additionally the country, as well as the world, experienced steep hikes in oil
prices during this period. All these situations, combined with the increased market liquidity, resulted in
strong inflationary pressures: by the late 1970s Paraguay experienced again double-digit inflation, which
reached 28.2% in 197915.
2.3. The Lost Decade of the Eighties.
As for most countries in Latin America, for Paraguay the 1980s were years of macroeconomic instability
and stagnation. Much of the investments carried out with the transitory resources flowing into the country
during Itaipú´s construction were not invested prudently and, thus, did not provide a buffer for the
coming letdown. Not surprisingly, the economic boom of the seventies ended in 1981, after which the
country suffered a recession period that lasted 2 years. Real GDP declined by 1% in 1982 and by 2% in
1983, while unemployment soar from 3.5% in 1981, to 7% in 1982 and to 12% in 1983. The economy’s
absolute contraction stopped in 1984, however growth remained at less than 2.5% per year for the next
three years, well below the 3% population increase per annum. In 1987 and 1988 recovery took place,
with growth even exceeding 6% thanks to the agricultural sector. However this was not enough to outset
the poor performance of the earlier years. By 1990, GDP per capita was 1.7% lower than in 1980.
Incidentally, the sectors that suffered most in the post-Itaipú era were those that had grown most rapidly
in the 1970´s boom. For example, although commerce and finance was 17% higher in 1989 than in 1981,
this sector average growth rate per year was below that of GDP´s during the eighties. The construction
sector was severely affected also, contracting more than 6% in 1982 and sustaining negative growth rates
in each of the subsequent years until 1986. As a consequence, construction activity in 1989 was still more
than 7% below that of 1981. There were hopes that Yacyretá (a second hydroelectric project planned with
Argentina) could counter this trend, but several problems resulted in a number of postponements of the
project. Others sectors did not perform well either as manufacturing output moved at about the same pace
as global GDP. Basic services – electricity, water and sewage and transport – expanded faster than the
14
Many private sector investments later proved to be over dimensioned, construction investment certainly was greatly exaggerated.
These rates of inflation were comparatively low for Latin America standards, a fact mainly explained by the extreme openness of
the economy and by the overvaluation of the local currency.
15
economy, but growth also declined substantially compared to earlier years. Even though it averages 8.5%
in 1987-89, the agricultural sector growth was still half the rate observed during the previous decade16.
A second factor contributing to the country’s stagnation was the world recession that particularly affected
Argentina and Brazil, Paraguay’s largest trading partners. These countries were themselves caught up in a
period of structural adjustment that forced them to reduce imports sharply and to devalue their respective
currency and, accordingly, Paraguayan exports began to decline. Worldwide recession also caused a
decline in the international prices of the main export’s products. For example, soybeans prices decrease
2.6% in 1981 and 5.4% in 1982, while cotton prices decrease 11.4% in 1981 and 16.7% in 198217.
TABLE 5: Registered Export Commodity Composition (%)
Products
1960
1970
1980
1990
2000
Wood Products
18.7
19.7
21.4
3.9
8.6
Livestock and meat
26.5
23.8
0.3
13.9
8.1
Tobacco
5.9
9.0
3.3
0.6
0.4
Cotton
1.1
6.3
34.1
34.7
10.6
Soybeans
0.0
0.0
13.6
27.9
32.9
Vegetable Oil
5.7
10.9
5.5
1.4
4.8
Quebracho extract
10.9
3.0
1.4
0.6
0.0
Others
31.2
27.3
20.4
17.0
34.6
Source: Central Bank of Paraguay.
The increasing external imbalance and the substantial reduction in aggregated demand presented a very
grim scenario for the Paraguayan economy. As a result, the government’s macroeconomic policy was
directed to avert the post-Itaipú recession through an investment and spending program. The government
began then to speed up the disbursement of foreign loans it had contracted in previous years and it
arranged new ones. Most of this increase in the external debt was due to public sector borrowing to
finance infrastructure as well as some heavy industries, like steel and cement.
However, the outcome of these actions differed substantially from expectations: investments turned out to
be unprofitable and over dimensioned for the domestic market. It targeted to sectors where regional
markets already showed substantial excess capacity. On the other hand, foreign debt increased
dramatically, rising from 15% of GDP in 1981 to 62% in 1987. In the mid-1980s, the government
16
It could be observed however a significant restructuring of output. This was now more oriented to the domestic market with an
impressive expansion in corn and wheat production.
17
Baer and Breuer (1986).
stopped servicing much of its international debt. As a consequence, the country lost access to
international markets and several creditors suspended the disbursement of already contracted loans.
In addition to the external borrowing, internal credit expanded quickly during this decade. Domestic
credit to the public sector (including the Central Bank’s deficit) increased sharply contributing to the rise
in prices18. Inflation, which had gone from 14% in 1981 down to 6.8% in 1982, rose and doubled in 1983
and continued to rise in all the following years. The implications of higher inflation were amplified by the
multiple exchange rate system instituted in 198219. Exporters were forced to surrender part of their
foreign exchange earnings to the Central Bank at below market prices; the Central Bank then sold foreign
exchange to favored buyers at a still lower price. The system generated a large Central Bank deficit that
had to be financed by inflationary monetary emission, distorted incentives against exports and gave
opportunity for corruption and easy profits for private and public official, the more so as the domestic
price level increased. The principal legal beneficiary of the multiple exchange rate system was the nonfinancial public sector. Although official figures suggest the non-financial public sector remained roughly
constant throughout the eighties, the explicit and implicit subsidies from the Central Bank and exporters
to the sector reached an equivalent of about 6% of GDP annually in 1986-88.
2.4. The Worst Decade: the Nineties.
As mentioned before, economic policy in the latter half of the 1970s did not prepare the country for postItaipú conditions and most of the transitory additional resources were spent as if they were permanent. On
top of that, the slow-down was aggravated by poor macroeconomic policy and rent seeking practices
during the eighties. In part the increased rent seeking practices were a sign of the crony authoritarian
regime of Stroessner. That regime, one of the longest lasting in the recent history of Latin America, was
weakening rapidly and it was toppled by the social unrest burst in February 1989. As expected after a
dictatorship of 35 years, the country embarked into a pseudo-democratic period characterized by a
wandering and unstable balance of power, that hopefully will converge to strong democratic institutions.
Needless to say, the political uncertainty of the period could have affected the incentives to invest in the
country. But, in general, many important changes in the appropriate directions have been observed during
the post dictatorship period.
18
For example, in 1978-84 domestic credit grew at an annual average rate of 26.5%.
The exchange rate was fixed at G. 126 to the US dollar since 1961. In mid-1982 a system of multiple exchange rates was
introduced, with rates ranging from 128 to 240 Guaraníes to the US dollar, with 160 being the dominant figure.
19
The authorities that took charge in 1989 substantially changed macroeconomic policy management. This
began almost immediately with perhaps the most important reform: the unification of the multiple
exchange rates. At the same time, all foreign exchange controls were removed and commercial banks
were permitted to deal in foreign exchange. This measure eliminated significant distortions as well as
Central Bank losses on foreign exchange transactions, eroding one of the major sources of inflation.
Moreover, the Guaraní was allowed to float, which resulted in a 92% nominal and a 24% real
depreciation.
TABLE 6: Main Macroeconomic Indicators
Indicators
1960
1970
1980
1990
2000
2.9
4.8
8.8
3.1
2.0
30.3
3.4
13.1
21.7
13.4
131
133
136
1,230
3,507
Int. Reserves, US$ Mill.
0.91
17.3
748.7
675.0
771.9
External Debt, US$ Mill.
26.7
146.9
690.6
1,669.9
2,234.3
Registered Exports, US$ Mill.
26.9
64.1
310.2
958.7
869.4
Registered Imports, US$ Mill.
32.5
63.8
517.1
1,193.4
2,050.4
Fixed Investment, % GDP
7.4
12.16
26.8
21.9
18.3
Tax Revenues, % GDP
n.a.
10.3
8.1
9.2
10.3
Public Sector Deficit, % GDP 1/
n.a.
n.a.
1.2
3.1
-0.9
GDP Growth, % per year
Inflation, % per year
1/
Exchange Rate, Gs/USD
1/
2/
1/
2/
Ten year averages: 51-60, 61-70, 71-80, 81-90, and 91-00.
Free market price. End of period.
Source: Central Bank of Paraguay and World Bank.
Thereafter, significant changes were made in public finances, international economic policy and financial
sector policy20. In the area of public finances, the public sector deficit was initially reduced, financial
management of the public enterprises tightened and public investment slowed. In December 1991 a new
tax code was passed simplifying and modernizing the tax system. In particular, the new tax code placed
greater reliance on indirect taxation, especially on the value added tax. Additionally, the elimination of
the debt to Brazil and the buyback of much Paraguayan debt to commercial banks reduced the country’s
interest burden and its susceptibility to external interest rate shocks21. In the financial sector, interest rates
were liberalized in 1990 and by 1991 they were completely freed. They were influenced only by the
Central Bank via the discount rate. Also, selective credit controls were abolished almost completely and
20
One of the main economic problem in previous periods was the large, but hidden, public deficit which attained around 8% of GDP
in 1988 (including interest arrears and the public sector foreign exchange subsidy). The same year official figures showed only a 3.1%
of GDP deficit.
21
In 1989 Paraguay reduced its debt with Brazil by over USD 400 million through a swap for Brazilian debt that it purchased in the
secondary market. In 1992, arrears to commercial bank creditors were reduced from over USD 200 millions to less than USD 3
millions, through the buyback of USD 172 millions of debt and the restructuring of most of the remaining debt.
reserve requirements reduced. With regards to trade policy, a new tariff code was passed in 1992,
lowering and simplifying tariffs to bring them in line with the de facto openness of the economy22. Also,
in 1991, Paraguay joined the common market, MERCOSUR, by signing the Treaty of Asunción along
with Argentina, Brazil and Uruguay. Since then, the country has complied with its obligations to reduce
tariffs to its MERCOSUR partners.
Nonetheless all the reforms taken, economic growth was far from outstanding during the first part of the
decade. GDP grew in average at 3.2% per year between 1990 and 1995, barely keeping pace with that of
the population. GDP continued heavily influenced by agricultural output, which were severely affected
by weather conditions in 1990 and 1991, further deteriorating in 1992. The next year showed a good
performance in the agricultural sector leading to a GDP growth rate above 4%. Regarding other sectors
in the economy, they mostly grew around 3% per year in average with the exception of the basic services
sector. This sector was the fastest expanded one with an annual growth rate above 7% in average,
reflecting the rural electrification program established by the electricity company and the investment
programs to increase water coverage in Asunción.
Inflation was usually kept under control during this period, except at the beginning of the decade. The
liberalization of the exchange in 1989, the adjustments on the prices of public services and a high rate of
monetary expansion explained by the accumulation of foreign reserves were the factors behind an
inflation rate of above 40% in 199023. Ever since, the exchange rate was used as a nominal anchor.
Initially the government sought to accumulated international reserves and to protect export
competitiveness, but then it began to worry more about inflation and less about exports, so that the
management of the exchange rate gave priority to price stability. Consequently, the exchange rate was
devalued at a rate below inflation and the local currency started to appreciate in real terms. At the end,
inflation was gradually lowered from 44.1% in 1990 to 13.4% in 1995, but this policy was held only until
the banking crisis in 1995.
On the whole, the initial reforms of the financial sector were undertaken in 1989 and 1990 and were
aimed at introducing a market-based system of monetary management. The underdeveloped and heavily
regulated financial sector was suddenly deregulated in an environment characterized by implicit deposit
insurance and relatively good, but weakly enforced, prudential regulations. This was compounded with a
slight appreciation of the real exchange rate during the nominal anchor period, which contributed to a
rapid consumption increase that resulted in excessive risk taking by the financial institutions. As a result,
22
External tariff rates went from a range of 3% to 86% prior to the reform, to 9.6% for manufactures, 6.9% for agriculture and 3.4%
for mining and quarrying.
a full-fledged banking crisis erupted in 1995. A second round of bank failures followed this in 199724. At
the end a total of 13 domestic banks (about a third of the banking system) and a number of other financial
institutions failed since 1995. The cumulative cost of the banking crisis during the period 1995-98 has
been estimated at around 13% of GDP.
During the second part of the nineties, monetary conditions reflected the increasing financing needs of
banks in distress. From the middle of 1995 to the end of 1997, the Central Bank was forced to absorb the
costs of the crisis. In order to mitigate the inflationary impact of the credits awarded to troubled financial
institutions, monetary growth was reduced. The exchange rate was defended by selling foreign reserves,
as rehabilitation credits were sterilized. In December 1997, after reserves had declined by close to 40%
from their peak in 1996, the Central Bank abandoned its support of the Guaraní, which depreciated by
some 20%. The weakening of domestic economic activity kept inflation below the rate of depreciation
and, in real effective terms, the Guaraní depreciated during these years25.
In the fiscal area, structural reforms and a comprehensive adjustment in 1990 led to a series of fiscal
surpluses during the first half of the decade that helped reduce the country’s debt burden. For several
years prior to 1997, the non-financial public sector achieved surpluses averaging around 2% of GDP a
year26. Beginning in 1997, however, the balance slipped into a deficit caused mainly by weakened tax
collection, strong wage growth and the need to undertake long delayed maintenance investment. Public
enterprises saw also their cash flow squeezed, as tariffs were not adjusted in line with costs. In addition,
the social security system lost its surplus position after half of its assets were frozen in intervened banks
and ceased to earn interest. The public sector deficit reached more than 5.5% of GDP in 2000, when
spending surged, financed by a large injection of external debt27.
As mentioned before, economic growth was moderate in the early years of the nineties and accelerated
briefly during a credit expansion in the mid-1990s. However, the banking crisis brought the expansion to
a halt in 1996 and, since then, the Paraguayan economy has undergone a sustained contraction. As a
result of the problems in the financial system, private sector credit contracted sharply and real interest
23
From March 1989 until end 1991, the Central Bank accumulated more than USD 700 millions in foreign reserves, equivalent to
about 20% of GDP. Most of these resources were later used to solve Paraguay’s external debt problems.
24
Many actors share the blame for the crisis: some bankers were inexperienced, a few were dishonest, some auditors were
incompetent, the Superintendency was impotent and the public conveniently turned a blind eye, assuming the government would not
leave them out in the cold if a fire erupted. (World Bank, 1999).
25
Inflation initially picked up to 14.6% at the end of 1998, but was rapidly reduced to single digit inflation the following year.
26
The surpluses were mainly explained by the increase in revenues. For example, the Central Government revenues increases from
about 8% of GDP in the late 1980s to about 15% in 1997. Most of the increase in revenues was due to import taxes, imposed on
triangular trade, and the introduction of a value added tax of 10% in 1992. Nevertheless, most of the revenue increases were spent on
personnel, with these expenditures rising from 3% of GDP in 1990-92 to about 7% of GDP in 1997.
27
Paraguay’s public external debt has doubled over the last five years to about 32% of GDP. In 1999, the country obtained a loan
from Taiwan amounting USD 400 millions, equivalent to 30% of the external debt at the moment. The resources were fully spent in
1999 and 2000 in a worthless attempt by the government to revert the economic stagnation.
rates rose to around 25%, while depositors increasingly shifted towards USD denominated assets.
Additionally, the recession led to a sharp rise in non-performing loans and increased bank’s reluctance to
extend credits to the private sector. Increasing capital flight during recent years reflected a deep lack of
confidence and compounded the shortage of credit in the financial sector. The steep decline in
investment, falling terms of trade, slow growth among trading partners and a contraction of the re-export
business combine to reduced real GDP first in per capita terms and, in 1998 and 2000, in absolute
terms28. As a consequence, poverty has deepened – especially in rural areas – and the income distribution
became more unequal.
3. Poverty, inequality and social indicators in Paraguay.
The previous section showed that, during the past 50 years, the economic performance of Paraguay has
gone trough different stages. The 1960´s were years of increased growth and financial and political
stability compared to the previous decades, with commerce and construction being the main sources of
growth. In the 1970´s Paraguay had a good economic performance thanks to agricultural expansion and
the construction of large hydroelectric projects. After the completion of the main works at Itaipú in 1981,
the Paraguayan economy entered a deep recession that lasted well into the mid 1980´s. Economic
recovery started in the first half of the 1990´s but a combination of factors, both domestic and external,
led to a new recession over the second half of this decade. A glance at the social indicators available over
this whole period shows that, in spite of some improvement over time in many of them, the bulk of the
Paraguayan population continued to live on a fairly low standard.
In this part of the paper we provide a brief summary of poverty, income distribution, health and education
environment in Paraguay. Data in some important economic and social indicators became available only
in recent years. For example, the household survey program in Paraguay started in 1983, and data have
been collected once every year since. However, until 1993 the surveys covered only the metropolitan area
of Asuncion and, just in 1995, the first national survey was implemented. All these issues have placed
important constraints on the analysis of this section.
The share of the population living in poverty is still high in Paraguay, both in the metropolitan area and in
the country as a whole. According to Robles (1999), the population in poverty in the metropolitan area of
Asuncion has been steadily decreasing between 1983 and 1997. In 1983, the share of the population in
poverty was 55.4% (of which, 16.2% were in the extreme poverty category). By 1990, this share was
41.6% (11.9% in extreme poverty) and by 1997 it further decreased to 23.7% (4.0%). At the same time,
28
The average growth rate of real GDP was only 0.7% per year in the period 1996-2000.
this author noticed that inequality, as measured by the Gini index, has remain stable in the metropolitan
area during most of the period, but slightly increased in the most recent years. The Gini index has a value
of 0.473 in 1983, 0.449 in 1987, 0.445 in 1990, 0.472 in 1993, 0.503 in 1995 and 0.483 in 1997.
TABLE 7: Poverty and inequality indicators in the metropolitan area of Asuncion
1983
1987
1990
1993
1995
1997/8
Extreme poverty
16.2
16.0
11.9
11.2
6.0
4.0
Poverty
39.2
32.2
29.7
24.8
21.5
19.7
0.473
0.449
0.445
0.472
0.503
0.483
Gini index
Source: Robles (1999).
Complementary, the World Bank (2001) found that, between 1995 and 1999, the share of the population
in poverty increased from 30.3 to 33.7 percent. At the same time, the share of the population in extreme
poverty increased from 13.9 to 15.5 percent. This increase in poverty was largely due to the country’s bad
economic performance. A severe banking crisis, falling agricultural prices and other shocks have
combined to produce the worst macroeconomic performance in all Latin America, with four recession
years in a row and an overall contraction of per capita GDP of 6.5 percent. Over this period, the number
of the extreme poor increased by 75,000 and the number of the poor by 180,000. The increase has been
especially large in rural areas. The households living in smaller cities and specially those living in rural
areas were more affected by the economic downturn than the households living in Asuncion30. Today,
although rural areas account for less than half the country’s population, they account for almost 80
percent of the extreme poor and 57 percent of the poor.
TABLE 8: Share of the population in poverty and extreme poverty
1995
1996
1997/98
1999
Country
13.9
n.a.
17.3
15.5
Urban Areas
6.8
4.9
7.3
6.1
Rural Areas
21.4
n.a.
28.9
26.5
Country
30.3
n.a.
32.1
33.7
Urban Areas
23.7
21.2
23.1
26.7
Extreme poverty
Poverty
30
According to some studies, rural poverty may have been rising all the way back to 1980, even during relatively prosperous periods
for the Paraguayan economy a whole, and even when urban poverty was declining. These previous studies suggest that, between 1983
and 1992, extreme poverty fell in Asunción whereas over similar period (1980-1992), it rose in the countryside.
Rural Areas
37.2
n.a.
42.5
42.0
Source: World Bank (2001).
The World Bank also estimates that inequality has increased during recent periods, contributing to high
rates of poverty. A common wisdom held in the eighties was that Paraguay had an equitable distribution
of income (due to a relatively low level of inequality in the metropolitan area of Asuncion). However,
when national survey data became available, the country emerged as one with the highest income
inequality in Latin America. Inequality at the national level, as measured by the Gini index, has increased
further in the second half of the 1990s because of an increase in rural areas, while it remained stable in
urban areas. Decompositions of inequality measures according to household characteristics suggest large
differences in income by education level, geographic location, and economic activity. According to the
World Bank (2001), geographical location, education and employment are found to account for at least
one fifth of national inequality. Household size matters less. While these findings are not surprising (they
have been observed in many other countries), the important message of these results is that the
contribution of disparities in education to inequality in income has been apparently increasing over time.
TABLE 9: Gini index of income inequality
1995
1996
1997/98
1999
National
0.581
n.a.
0.592
0.597
Urban areas
0.515
0.485
0.502
0.497
0.476
0.476
0.451
0.472
0.563
n.a.
0.609
0.664
Metropolitan Asunción
Rural Areas
Source: World Bank (2001).
While there has been virtually no increase in per capita GDP over the last two decades, Paraguay has
shown progress in some non-monetary indicators. For example, the UNDP´s Human Development Index
(HDI) for Paraguay improved from 0.663 in 1975 to 0.738 in 1999, but it remains below the level
achieved by most other Latin America countries. The improvement in non-monetary indicators despite
lackluster growth may be surprising, yet it could be due to higher public sector social spending and
urbanization. However, Paraguay´s progress in the HDI (an increase of 0.075 over 25 years), while
similar to the improvements observed in Argentina and Uruguay, have been below those observed in
countries more comparable to Paraguay such as Bolivia, Chile and Colombia.
TABLE 10: Level and trend in the Human Development Index (HDI).
Paraguay
Uruguay
Chile
Bolivia
Argentina
Colombia
1975
0.663
0.755
0.700
0.512
0.784
0.657
1980
0.698
0.755
0.735
0.546
0.798
0.686
1985
0.704
0.779
0.752
0.572
0.804
0.700
1990
0.716
0.800
0.779
0.596
0.807
0.720
1995
0.733
0.813
0.809
0.628
0.829
0.746
1999
0.738
0.828
0.825
0.648
0.842
0.765
Source: World Bank (2001).
Paraguay has also made substantial progress in educating its labor force over the last decades, but it still
lags behind for enrollment in secondary education. According to the United Nations Development
Program (UNDP), illiteracy, in the population aged 15 years and over, decreased from a high 34,2% in
1950 to 22,8% in 1982. It was further reduced to 9.7% in 1992 and to 8.4% in 2000/01. Also, as
indicated in Table 12, the number of years of schooling for the population aged 15 and over has
increased gradually since 1970, but it barely pass 6 years in 2000. That is, Paraguay still lags behind for
enrollment in secondary education. This suggests that there is a low transition from primary to secondary
school, which is one of the issues that the education reform has addressed by expanding the primary
cycle to nine years instead of six31. The same table shows that the number of years of schooling in
Paraguay is low when compared to other countries in the region. The relatively low level of the
population’s human capital, as measured by the number of years of schooling, could be an important
factor in explaining the meager economic growth rates observed in Paraguay in recent years.
TABLE 11: Trend in illiteracy rate and net enrollment rates
Illiteracy rate
1/
1/
1960
1970
1980
1985
1990
1997
25.6
20.0
22.8
n.a.
9.7
8.4
19.0
15.2
20.0
n.a.
8.3
6.9
1/
31.5
24.8
25.5
n.a.
11.8
9.8
Enrollment primary
n.a.
n.a.
88.7
89.5
92.8
91.2
Enrollment secondary
n.a.
n.a.
n.a.
25.4
25.8
37.9
Men
Women
1/
Years 1962, 1972, 1982, 1992 and 2000.
Source: UNDP (2003) and World Bank (2001).
31
Returns to education in Paraguay are similar to those found in other countries in the region. Psacharopoulos (1994) found that the
coefficient on years of schooling in a mincerian regression was 11,5% in Paraguay, 10,3% in Argentina, 14,7% in Brazil, 12,0% in
Chile and 9,7% in Uruguay.
Regarding the quality of education, multilateral agencies found that education efficiency is low. The
massive recruitment of primary school teachers, with low proportions of certified ones, to face the
increasing demand for education during the nineties, negatively affected the quality of the education
system. In 1997, only 59.1 percent of teachers held the proper academic qualification to teach while 66.7
percent were certified to teach.
TABLE 12: Average years of schooling in the population aged 15 years and over
Country
1970
1980
1990
2000
Argentina
6.2
7.0
8.1
8.8
Brazil
3.3
3.1
4.0
4.9
Chile
5.7
6.4
7.0
7.6
Paraguay
4.2
5.1
6.1
6.2
Uruguay
5.7
6.2
7.1
7.6
Source: UNDP (2001).
Paraguay’s fertility rate (and consequently, its rate of population growth) is among the highest in Latin
America, even though it has been decreasing over time. The country´s fertility rate was 6.6 in the period
1960-65, 5.7 in 1970-75, 5.3 in 1980-85, 4.6 in 1990-95 and 4.2 in 1995-00. The average fertility rate for
the countries in Latin America and the Caribbean during the same periods were 6.0, 5.1, 3.9, 3.0 and
2.750. The World Bank (2001) also found that fertility rates have decreased in the 1990s, but with a
leveling off in recent years. The fertility rate has diminished from 4.6 in 1987-1990 to 4.3 in 1990-95,
and has remained stable thereafter. However, fertility rates in rural areas are still 60 percent higher than in
urban areas: for the same periods, fertility rates in rural areas diminished from 6.0 to 5.6, while in the
urban area it declined from 3.6 to 3.2. At the same time, life expectancy of the population have been
increasing. According to the UNDP (2003), in 1950 the life expectancy of a typical Paraguayan was only
62.7 years. In 1960 it was 63.8 years, in 1980 it reached 65 years, in 1990 it was 67.1 years, while at the
50
See Economic Commission for Latin America and the Caribbean (1981) and (2001).
end of 2000 it attained 70.1 years. Both factors have put pressures on the yearly rate of economic growth
necessary to rise per capita income in the country.
Finally, it must be mentioned that while there have been no increase in per capita GDP and probably no
decrease in poverty over the last two decades, there has been a reduction over time in unmet basic
needs51. The share of households with at least one unmet basic need has decreased by about 30
percentage points nationally and in urban and rural areas. For example, the share of household with at
least one unmet basic need dropped from 86.9 percent in 1982 to 55.3 percent in 1997/98. Similarly, this
share declined in the same period from 72.2 percent (99.5) to 44.7 percent (67.5) in the urban (rural)
area.
4. Accounting for Growth.
In this section we use a production function approach to determine the sources of growth in Paraguay
during the period 1962-2000. The objective is to separate the roles of input accumulation and that of the
increase in total factor productivity (TFP) in the growth of aggregate output. For reasons that will become
apparent soon, we employ a variety of methodologies to make the decomposition.
First we explore the behavior of the relevant time series. Figure 1 displays the series of output, capital,
employment, labor quality and cultivated land. All the series are expressed as a ratio of their original
value in 1962, with the exception of cultivated land which, for data limitations, is expressed as a ratio of
its value in 1966. While it is clear that all series have been increasing consistently over time, their
behavior is very different. More importantly, as discussed before, the rate of growth of per capita output
is upsettingly low.
The series of capital is estimated using the traditional perpetual inventory method with a depreciation rate
of 8% per year. We find that for almost any relevant value of the depreciation rate, the implied series of
capital would grow much faster than aggregate output. Thus, we employ this rather high (but still
reasonable) value for the depreciation rate to minimize the implication of a possibly artificial
capital/output ratio on the growth accounting exercises. However, the qualitative conclusions are robust
to the use of depreciation rates of 4% or 6%. Employing those rates would certainly change the numbers,
but not dramatically. Indeed, we will make the case later that the rising capital/output ratio is simply a
reflection of the poor behavior of total favor productivity.
51
Paraguay´s index of unmet basic needs uses six indicators: water, sanitary installation, primary education, subsistence capacity,
crowding and housing material.
Figure 1: Basic Aggregate Time Series of the Paraguayan Economy
We constructed the employment series based on the series on population and participation rates by gender
from the World Bank dataset. Results do not change much if instead we simply use the series in the
Summers Heston’s Penn World Table. In the paper We also consider cultivated land as another input in
production. The numbers of hectares are those recorded in Cabello et al (2000). We additionally
investigate a series of employment corrected for labor quality. Such correction is based on data on the
educational attainment of the labor force and differences in salaries and wages of different education
groups. However, those corrections had negligible effects, which is consistent with the results on the
poverty and education statistics discussed in the previous section.
Several features are evident from Figure 1. With the exception of human capital, all the series grow over
time. But they differ greatly in their behavior over time and across different periods of time. For example,
there is a high growth of output during the 1970s, a sharp recession in the early 1980s, a timid recovery
during the early 1990s and a subsequent decline. Capital growth also slows down after the 1980s, but it is
clear that the capital/output ratio has increased ever since.
During the sample period, cultivated land grew significantly also. Indeed, between 1975 and 1980 it
almost doubled. Yet, from 1980 to 1999 it grew only by an accumulated 20%, reflecting the fact that the
country is reaching the limits of the potential cultivable land.
The most striking feature is the lack of growth in the quality of labor. It is natural that this index cannot
grow at the same rate as other aggregate series. But the Paraguayan case is actually very disappointing as
not only the country started with one of the lowest index of human capital in the region, but moreover, its
improvements lagged behind with respect to other Latin American countries. There are theoretical and
empirical reasons to believe that human capital is one of the most important factors behind growth. Thus,
the lack of human capital accumulation becomes a prime suspect in the particular case of the meager
growth rates of Paraguay.
Other noticeable feature that one observes in the data is the disparate growth in the stock of constructions.
Many observers in the country believe that the over investment in housing is widespread in Paraguay.
Given this, we will explore econometric specifications and growth accounting techniques that consider
investment on equipment and on constructions as separate. Figure 2 reports the behavior of total capital,
capital in equipment and capital in constructions (structures) with respect to GDP. The series are
normalized to be equal to one at the beginning of the period. All those items grow at a faster pace than
output, specially constructions after mid 1970s. As can be seen clearly from the figure, the stock of
constructions/GDP rises sharply in the second part of the 1970s and also in the first part of the 1980s,
reflecting in part the push from the Itaipú project. But the upward trend is also present in the stock of
equipment/GDP. This ratio increases by more than 50% during the sample period.
The only way to revert these results is to use incredibly large depreciation rates. We believe, however,
that these rising capital/output ratios are the reflection of a declining total factor productivity. We next
discuss a battery of methods to decompose factor accumulation and show their results.
Figure 2: Capital Output Ratio Using Different Measures of Capital
4.1 Input accumulation and total factor productivity.
Given the time series of output and different inputs, the natural question is, what is the quantitative
contribution of each input in output growth and how much is attributable to an overall increase in
productivity?. We will examine several different methodologies to do this accounting exercise. First, we
use simple econometric methods to estimate a production function. Then, we use the point estimates and
the residuals of the regression to compute the contribution of each factor and that of the TFP. In the
second method, we use the parameter values traditionally used in the literature (and estimated from data
on other countries).
Our third decomposition is more innovative and was suggested to the authors by Rodi Manuelli. Here we
abandon the Cobb-Douglas assumption and allow for a Constant Elasticity of Substitution (CES)
production function. Using information on the share of output that goes to labor, we can estimate the
substitution and distribution parameters of the production function. Then the residuals of the regression
can be used to estimate a stochastic process that dictates the relative improvement in the units of capital
with respect to labor. With those at hand, we could estimate a TFP-like factor and separate the
contribution of factors and productivity in the behavior of growth.
The three methods are wildly different and produce very different quantitative results. However, all of
them point to a declining TFP as a main explanation for the poor growth performance of Paraguay.
The first method is based on the estimation of a Cobb-Douglas production function:
Y ( t ) = A ( t ) K ( t )α L ( t ) β
Here Y(t) indicates aggregate output, A(t) is the total factor productivity, K(t) the flow of services from
capital (which are assumed to be proportional to the existent capital) and L(t) the flow of services from
labor. Taking logs
y(t ) = a(t ) + αk (t ) + βl(t )
where lower case variables indicate the natural log of that variable in capital letter. This equation can be
estimated directly using Ordinary Least Square (OLS)52. However, there is another way to estimate the
unknown parameters in the production function. That is using the “intensive” form
[ y ( t ) − l ( t ) ] = a ( t ) + α [k ( t ) − l ( t ) ]
This specification imposes constant returns to scale in K(t) and L(t) .
Also, in earlier sections of the paper we explored an extension using cultivated land, i.e. one that assumes
that output is given by
Y ( t ) = A ( t ) K ( t )α L ( t ) β T ( t ) χ
where T(t) denotes cultivated land. However, the results we obtained were not interesting and point
estimates on land were close to zero. This is probably due to the fact that the period in which cultivated
land grew the fastest was precisely when capital was also growing very rapidly and even faster than land.
52
However, there are severe econometric problems with this approach (see below).
An extension that we do find interesting is the separation of capital between constructions and equipment.
That is, we explore an aggregate production functions of the form
Y ( t ) = A ( t ) Kb ( t ) α Ke ( t ) χ L ( t ) β
where Kb and Ke stand for the stock of capital in constructions and in equipment, respectively. We also
explore correcting the series for the “quality” of capital and labor. However, we have serious reservations
on the potential quality of the correction of the physical capital series. Information on interest rates and/or
price of capital, which are of critical importance for those corrections, is very distorted and fragmented in
a country like Paraguay, with a long history of government intervention in the financial markets.
Additionally, we shall assume that the log of total factor productivity follows a trend stationary process
of the form:
a ( t ) = a 0 + a1t + u ( t )
where u(t) is a random disturbance. With consistent estimates of a0 and a1 and given the value of u(t), the
estimated TPF of Paraguay at time t is given by
A ( t ) = exp[ a 0 + a1 t + u ( t )]
Depending on the stochastic properties of u(t), the standard error of the estimated coefficients would need
to be corrected. However, despite their popularity, econometric estimations of production functions of the
previous forms have a fundamental limitation. This has to be emphasized. All regression estimates of the
coefficients of production functions require the orthogonality of the residuals with the regressors. But,
economic theory indicates that the amount of labor L(t), and of capital K(t) -to the extend that it can be
adjusted in the short term-, must respond to the value of the residual. Thus, regression estimates are
inconsistent. This fact has to be bore in mind whenever interpreting the results.
Table 13 reports the results of all the different specifications, for both, intensive and extensive forms.
The t-statistics reported in the table are computed simply using the OLS standard errors. Of course, there
are valid concerns on the relevance of these standard errors. But there are econometric problems of
fundamentally higher relevance than obtaining robust errors.
First, as we indicated above, economic theory strongly indicates that the regressors cannot be orthogonal
to the residual. Periods of high TFP are also periods when investment is more profitable. Thus, the OLS
estimates are inconsistent. But, and perhaps more importantly, the table shows strong anomalies with the
obtained estimates with respect to the literature. For example, many times the results explicitly or
implicitly yield negative point estimates for one of the inputs. This happens with more strength for labor
and for capital in constructions.
Table 13: Estimation Results for the Aggregate Production Function of Paraguay, 1962-1999*
Explanatory\Dependent
Constant
log. Labor
log. Total Capital
Log (Output)
Log (Output /Labor)
0.12
0.06
0.03
0.05
0.03
0.05
(2.47)
(4.27)
(1.60)
(1.07)
(1.00)
(1.90)
-0.01
0.77
0.18
(-0.01)
(6.14)
(3.77)
0.68
(0.96)
log. Capital Equipment
0.44
0.52
(11.44)
(11.89)
-0.60
log. Capital Constructions
(-4.92)
2.29
log. Total Capital/Labor
(4.51)
log. Cap. Equipment/Labor
0.65
0.62
(8.71)
(8.47)
0.28
log. Cap. Construct/Labor
(1.30)
Time Trend
R-square
-0.00
0.02
0.00
-0.04
-0.02
-0.02
(-0.22)
(5.04)
(1.36)
(-10.34)
(-8.15)
(-19.45)
0.987
0.99
0.997
0.92
0.960
0.958
* Small case numbers are the t-statistics under the null that the coefficient is zero.
Source: Authors´ estimates.
Using basic economics, it is clear that we cannot take seriously some of these results. For instance, in the
second column, the regression in extensive form yields a value for the share on labor that is on the ball
park of the literature, but it absurdly implies negative productivity of constructions. For the same reason
we also disregard the results for the first regression, as it implies a negative (yet negligible) productivity
of labor. We discard as well the third column because it implies decreasing returns to scale, a problem
that is not resolved by introducing land in the regression. Thus, all the extensive form regressions cannot
be used to decompose the sources of growth. Finally, we also ignore the fourth column, i.e. the first
regression in intensive form. It implies a huge productivity of capital and negative productivity of labor,
which obviously does not make any sense.
Consequently, the only usable results are in columns five and six. We will use these two regressions in
the growth accounting exercises below. We also employ a value of 66% for the labor share, a percentage
that is commonly used in the literature and that is obtained from data on other countries. We employ that
share considering capital as the sum of equipment and structures. Results are similar if instead we use
capital as compose solely by the stock of equipment. This variety of methods allows us to check for the
robustness of the results.
Before exploring the third method, it is important to notice a remarkable finding: in most regressions the
time trend consistently shows a negative and statistically significant coefficient. Also, the results show
that the implied behavior of TFP from the different econometric regressions is very similar to each other.
In Figure 3 below, we report the implied behavior of TFP from all the models considered.
The low average labor share makes us highly suspicious of measurement problems, specially of labor
remunerations. This is the main reason we also computed the TFP and did the growth accounting
exercises imposing the value of capital-output share to 1/3, which is the standard in business cycle and
growth applied general equilibrium literature. We also explored a similar exercise including land with a
capital share of 1/3 and several values for the land share, including values as high as 1/3. The exercises
including land provide very similar quantitative results because land expand the most during the Itaipú
period, which is also the period in which capital was growing the fastest. We included those results in a
previous version of the paper, but we omit them here for the sake of brevity.
Figure 3: Total Factor Productivity Using Different Methods for the Paraguayan Economy
The previous graph shows the measures of TFP implied by the methods discussed above. All series are
expressed as a ratio of the TFP in 1962. The different methodologies yield different behavior, specially in
terms of magnitudes. However, there are strong similarities among them. Most of them indicate that TFP
consistently grew in the 1960s and 1970s and then dramatically fell in the early 1980s. They also agree
that TFP fell in the late 1990s. However, they differ on the implied behavior for the late 1980s and early
1990s. While some methods indicate that TFP starts recovering in that period, others illustrate that it
keeps falling. But more importantly, the vast majority of the methods show that, from the beginning of
the period, there have been a declining (or at best stagnant) path for the total factor productivity of the
economy.
Instead of pursuing a refinement of the econometric estimation of the standard errors, we believe that
there are much higher returns to explore and compare alternative methods to extract the total factor
productivity in Paraguay. One common method is to directly look at the share of labor earnings on GDP
from the National Accounts. Given the constant returns to scale and Cobb-Douglas functional form
assumptions, one can use the average (or median) value during the sample period to estimate the labor
share. The following figure shows the behavior of the labor share on output for Paraguay during the
sample period.
Figure 4: Output Labor Share in the Paraguayan Economy
The graph vividly shows relevant features of the series. First, the share is unusually low with respect to
international evidence. The average value is only 34.56%. Second, the share is not stable over time. It
reaches almost 40% in the late 1970s, but it falls to less than 29% in 1990. Interestingly, the labor share is
highly pro-cyclical. It raises during the Itaipú episode, then it falls sharply during the 80s and recovers
during the early 90s. This behavior casts doubts on the Cobb-Douglas functional form, but it may also
reflect measurement problems, labor market frictions, capacity utilization, relative price fluctuations, etc.
In what follows, we explain how extending the model to a CES and using information on the labor share
can be used to obtain two technological factors: TFP and a relative measure of capital efficiency.
4.2. Separating TFP and Capital Augmenting Productivity Improvements.
Imagine that instead of the traditional Cobb-Douglas, the production function takes the form of a Constant
Elasticity of Substitution (CES), i.e. aggregate output is given by
y( t ) = B( t )[θ (λ l ( t ) L( t )) − ρ + (1 − θ )(λ K ( t ) K ( t )) − ρ ]−1 / ρ
where B(t) is the total factor productivity while λl (t ) and λ K (t ) are labor and capital augmenting
technologies. The latter are quality indexes which effectively act as if the total units of labor and capital
had increased. We can factor out one of these two and we opt to factor out the labor quality, obtaining:
y( t ) = B( t )λ L ( t )[θ ( L( t )) − ρ + (1 − θ )(λ K ( t ) / λ L ( t ) K ( t )) − ρ ]−1 / ρ
This expression simplifies to
y( t ) = A( t )[θ ( L( t )) − ρ + (1 − θ )(λ ( t ) K ( t )) − ρ ]−1 / ρ
where we define the two components of productivity improvement as total factor productivity,
A( t ) = B( t )λ L ( t )
and relative improvement of capital with respect to labor, i.e.
λ ( t ) = λ K (t ) / λ L (t )
Clearly, the first term is a total factor productivity term that incorporates the common improvement of
labor and capital quality.
Assuming that factor prices are competitive, then the share of output that goes to capital is
sk (t ) =
∂y( t ) K ( t )
= (1 − θ )[θ (λ ( t ) K ( t ) / L( t )) ρ + (1 − θ )]−1
∂K ( t ) y ( t )
After some easy manipulations, it yields
1 − sk (t )
θ
[λ ( t ) K ( t ) / L( t )] ρ
=
s k (t )
(1 − θ )
This equation can be estimated. In particular, taking logs,
 1 − sk (t ) 
 θ 
 = ln
 + ρ ln(K ( t ) / L( t )) + ρ ln(λ ( t ))
ln
 (1 − θ ) 
 sk (t ) 
We further assume that the relative capital/labor quality follows a trend stationary process of the form,
λ ( t ) = exp[ ρ λt + ε ( t )]
where ε(t) is a random process. Then we obtain the equation:
 1 − sk (t ) 
 θ 
 = ln
 + ρ ln(K ( t ) / L( t )) + ρ λt + ε ( t )
ln
 (1 − θ ) 
 sk (t ) 
Notice that this relationship should hold regardless of the behavior of the TFP, A(t), of the economy.
Running a simple regression of the form
 1 − sk (t ) 
 = α + β ln(K ( t ) / L( t )) + δt + ε ( t )
ln
 sk (t ) 
we could obtain estimates of α, β and δ. With those estimates, we could then calculate the parameters of
the production function as,
∧
θ=
exp(α )
1 + exp(α )
∧
ρ=β
λ =δ / β
With those estimates in hand, we can back out the implied value of λ (t ) = exp[ ρ λt + ε (t )] using the
residuals of the regression as consistent estimates of the shocks. Finally, we can compute the implied A(t)
as
A( t ) = y( t ) /[θ ( L( t )) − ρ + (1 − θ )(λ ( t ) K ( t )) − ρ ]−1 / ρ
The point estimates of the regression, using only capital equipment, are θ = 1.7357e-007 and ρ = -0.2491
with a R square of 38%. Very similar results hold if total capital is used. Thus, the data suggest that
capital and labor in Paraguay are more substitutable than what a Cobb-Douglas production function
suggests and, moreover, that the contribution of labor is negligible.
With those estimates, we computed the implied TFP and the relative efficiency of capital, which are
reproduced in the following figures.
Figure 5: Total Factor Productivity and Capital Augmenting Progress.
Using Capital in Equipment Only
Figure 6: Total Factor Productivity and Capital Augmenting Progress.
Using Total Capital
Notice that the implied behavior of total factor productivity is in line with the previous methods: we
observe a steady decline following 1970s in TFP. Moreover, there are periods in which TFP and the
relative efficiency of capital move in the same direction, specially in the period between mid 1970s up to
1990. In that period, both TFP and the relative quality of capital decline gradually. This period coincides
with Itaipú project and the lost decade of the 1980s. It is interesting however, that both behave very
different in the 1990s. The results suggest that the quality of the investment in capital had an important
recovery, while TFP was still declining.
4.3. Growth Accounting.
Table 14 shows the results of the growth in output decomposition, showing the determinants of output
growth in five different subperiods as well as in the whole sample period. The panel shows the percentage
accumulated growth in output, measured as log differences, and the contribution to output growth from
total factor productivity and from the accumulation of production inputs, using the parameters for the
labor share. The first two panels show the results for the methods employing the typical value of labor
share used in calibrations. The first uses capital as the sum of structures and equipment, while the second
employs capital as equipment only. The third panel do the same calculation, but using the coefficients
estimated from the regressions in the previous section. The fourth panel uses also the regression estimates,
but allowing structures and equipment separately.
The last panel uses the CES specification. Growth accounting in this case is not as straightforward and
requires more elaboration. Here it is not possible to decompose the contributions of the different
production inputs linearly. Accordingly, the sum of the contributions between factors and TFP across
subperiods need not to add up. What we report in the table is the one-factor contribution. That is, we
computed by how much output would have grown if only one of the factors were augmented by the value
at the end of the subperiod, with all the other factors remaining at the same values as in the beginning of
the period. For example, to calculate the contribution of labor on output growth between 1971 and 1980,
we compute the implied output with the labor of 1980 but with the capital, capital efficiency and total
factor productivity as of 1971. We then write down the log difference with the actual output of 1971. As
with the other cases, here we multiply those differences by one hundred to express them in percentage
terms.
The most striking feature in the table is that, indeed, TFP has had a negative contribution on output
growth in most of the periods. Basically all the methods point in that direction. Perhaps the most
dubitative in this conclusion is the method allowing for a CES production function. However, in this case
notice that in the sub-periods when TFP has a positive contribution, typically capital efficiency has a
significant negative contribution. In sum, the growth decomposition indicate that the productivity of
factors, much more that their accumulation, is a major negative factor in the growth process of Paraguay.
The exercises show also that in general physical capital had a strong pull for growth in most of the
periods, but all agree that the contribution is much higher for the period of 1971 to 1980, which is
precisely when the construction of Itaipú took place. Furthermore, all the methods agree that the
contribution of capital in the 1990s has been modest when compare to other periods. We believe that the
political uncertainties of the period have lessen the incentives to invest in physical capital in Paraguay.
However, over the entire sample period, it is clear that capital accumulation has outpaced output, so its
accumulation does not seem to be the major deterrent of growth. The main problems appear in the
accumulation of human capital and the overall productivity of factors. It seems safe to conclude that if
Paraguay is to grow faster, more aggressive policies should focus on these problems. Any contribution in
this respect would also have the benefit of fostering the accumulation of complementary factors, such as
physical capital .
Table 14: Productivity and Factor Accumulation Contributions to Output Growth (%)
Period
Ouput
1962-1970
1971-1980
1981-1990
1999-1991
1999-1962
36.5
78.5
22.0
19.8
170.7
Ouput
1962-1970
1971-1980
1981-1990
1999-1991
1999-1962
36.5
78.5
22.0
19.8
170.7
Ouput
1962-1970
1971-1980
1981-1990
1999-1991
1999-1962
1962-1970
1971-1980
1981-1990
1999-1991
1999-1962
Labor Share=2/3, all capital
TFP
Capital
7.2
13.4
-22.8
33.9
-25.7
17.4
-16.0
14.3
-64.8
86.5
Labor share=2/3, Equipment only
TFP
Capital
2.0
18.7
-19.5
30.6
-21.5
13.2
-6.2
4.5
-51.1
72.8
Regression, Equipment only
TFP
Capital
Labor
15.8
67.4
30.3
21.5
149.0
Labor
15.8
67.4
30.3
21.5
149.0
Labor
36.5
78.5
22.0
19.8
170.7
-7.3
34.8
9.0
-16.8
57.2
38.2
-19.8
24.6
17.1
-0.8
8.4
12.2
-49.6
136.0
84.4
Regression, Equipment and Structures Separated
Output
TFP
Equipment
Construction
Labor
36.5
-13.2
36.1
11.7
1.9
78.5
-15.7
59.3
27.0
7.9
22.0
-24.3
25.5
17.2
3.6
19.8
-5.0
8.7
13.6
2.5
170.7
-64.4
141.0
76.6
17.5
Regression, CES and Separation of TFP and Relative Capital Efficiency
Output
TFP
Labor
Capital Rel. Capital Eff.
1962-1970
36.5
1971-1980
78.5
1981-1990
22.0
1999-1991
19.8
1999-1962
170.7
Source: Authors´ estimates.
18.9
-30.0
1.9
-35.0
-54.1
0.0
0.0
0.0
-20.1
170.0
23.7
101.1
45.4
32.3
202.3
-6.1
7.4
-25.3
22.6
192.6
4.4. Discussion.
Without any doubt, the most salient finding is the behavior of TFP. Not only, it shows a weak growth
overall, but, more strikingly, there is strong evidence for a persistent decline during the eighties, and more
surprisingly during the nineties. The decline in the eighties is not really surprising in light of the results
for most Latin American countries. But, an open question is why TFP fell during the nineties?
The data available at this point is not very useful in addressing this issue. Hence, we can only speculate. A
possible explanation lies on the political uncertainty due to the transition from a dictatorship to an infant
democracy. The presence of labor and credit market rigidies would impede the reallocation of labor and
capital, and under increased uncertainty, the under-utilization of capital and labor, could be reflected in a
decline of TFP. Another part of the explanation would rely on the lack of reforms during the nineties. As
opposed to most Latin American countries, and specially the ones in the Southern Cone, Paraguay showed
very little progress in its privatizations and trade reforms. Paraguay has been less successful in attracting
foreign investments than the rest of the region. These factors could be underlying a departure of the
behavior of Paraguay with its neighbors. An additional force behind a decline in TFP could arise from the
disruption of the banking system during the nineties. The banking crisis observed during that period could
have prevented the allocating of capital to the most productive uses. Finally, the implementation fo
MERCOSUR, may have represented a negative TFP shock to the Paraguayan economy, as it eliminate the
role of Paraguay as the middle-man for the trade among its neighbors.
An entirely different set of explanations can be found on the fiscal incentives to over-accumulate capital
and incentives to over-report investment and/or under-report income. Such hypotheses find support in
cases like Itaipu, discussed above, and the fact that structures have been accumulated at a faster pace than
equipment in Paraguay. It is widely believed that equipment have a higher contribution to growth than
structures. It is also widely believed that there is a huge problem of income-tax evasion in Paraguay.
Finally, it would have been useful to have data on input prices, i..e. wages and rental rates of capital and
land.
In reserve of distorsions and rigidities, the behavior of wages and interest rates indicate how
productive labor and capital have been in Paraguay, and with the information on their quantities, one
could construct alternative series of productivity. Moreover, the behavior of relative wages, i.e. across
skill groups and industrial sectors, could be used to sharpen our analysis of aggregate productivity and its
composition across sectors. Furthermore, such information could be used to understand the degree of
complementarity or substitubility of capital with different forms of labor. This is important, as it is very
likely that the degree of capital-skill complementarity has changed significantly in the last years as the
result of regional integration agreements. But again, the lack of data is the main limitation.
5. Growth and Other Main Macro Variables.
In this section we investigate the relation of growth in Paraguay with domestic institutions and policies.
Empirical studies analyzing the long run determinants of growth typically relate the real per capita growth
rate to two kinds of variables. First, the initial levels of state variables, such as stocks of physical and
human capital. Second, environmental and policy variables such as the ratio of government consumption
to GDP, the ratio of domestic investment to GDP, movements in the terms of trade, inflation, measures of
political instability and the rule of law, tariff rates, and so on. These empirical studies usually employ
cross-country evidence (data) to identify what are the policy and institutional factors that are significantly
related to the growth rate of real GDP53. That is, the analysis is based on a general framework of crosscountry regressions, which puts the experience of individual country in a global context. Since these
regressions apply to a panel set of cross-country data over a few decades, the papers contain a limited
amount of time series variation.
Although cross-country data seems to support the hypothesis that several external environmental and
policy variables could affect output growth, it is important to test if these findings hold for a single
country over time. However, a test on a time series data for a single country could be difficult to carry out.
One element of the problem could be to obtain a suitable approach to defining the long run and detecting
long run relationships. In this part of the paper, the basic proposition that the growth rate of the economy
and some of the factors usually considered in growth regressions are correlated is examined from a nonstructural, low frequency point of view. This methodology is based on Lucas (1980). In this paper, Lucas
presents empirical illustrations of two central implications of the quantity theory of money: that a given
change in the rate of change in the quantity of money induces (i) an equal change in the rate of price
inflation; and, (ii) an equal change in nominal rates of interest. Since the two quantity-theoretic
propositions hold only in the “long run”, Lucas constructs a filter to smooth the original data (i.e. to
extract its long run components) before testing the implications of the theory.
Here, we use the approximate band-pass filter developed by Baxter and King (1995) to obtain the low
frequency components of the time series. For the empirical applications, we adopt the definition of the
business cycle suggested by the procedures and findings of NBER researchers, like Burns and Mitchell
(1946), that specified business cycles as the cyclical components between eighteen months and eight
years. We adopt these limits as the definition of the business cycles so, to isolate the trend or low
frequency of the data, we consider those frequencies with periodicity of eight years or higher.
Specifying the business cycle as fluctuations with a specified range of periodicity results in a particular
two-sided moving average (a linear filter). In the particular case of the NBER definition of the business
cycle, the desired filter is a band-pass filter, i.e., a filter which passes through components of the time
series with fluctuations of eight years or higher while removing components at higher frequencies.
However, the resulting moving average is of infinite order and an approximation to this filter is necessary
for it to be applicable to finite time series. Therefore, in order to analyze the hypothetical relationship in
the log run between economic growth and each of the factors considered, we first apply the following
filter to the original time series data:
y t* =
k
∑a
j =− k
j
yt + j
where y t* is the value of the filtered series. The optimal approximate filter weights, aj, are functions of
the weights of the ideal low-pass filter, bj ,and an adjustment term, θ. Thus, aj = bj +θ 54.
A parameter to be chosen is the value of k, the number of leads and lags in the filtered series. We have set
this value equal to six55. Thus the approximate band-pass used in this analysis is the BP6(8) filter
described in Baxter and King (1995). The notation reflects the fact that the filter passes through
components of the data with cycles higher than 8 years and the subscript "6" means that 6 leads and lags
of the data were used in constructing the filter (i.e. 6 annual observations are lost at the beginning and end
of the sample period for the filtered data).
A wide variety of external environment and policy and policy variables could affect growth rates by
changing the long run potential income and the rate of productivity growth. Based on the results from
previous empirical research, we consider the following variables as important determinants of long-run
per capita income: (1) inflation rate, (2) government consumption, (3) investment rate, (4) private
consumption, (5) exogenous shock (terms of trade growth), and (6) growth of exports.
1. Inflation rate: in recent years, the contours of an inverse connection between inflation and growth
across countries have begun to emerge from econometric studies. For example, Barro (1991) reports a
negative relationship between inflation and the growth rate of real GDP during 1970-1985 in a cross
section of 117 countries. Similarly, Fischer´s (1993) cross section regression estimates, based on data
53
See for example De Gregorio and Lee (1999) for a paper that examines the growth experience of Latin America countries or Barro
(1991) for a study the uses cross-country data from developing and developed countries.
54
For a more detailed discussion of the issues involved in constructing the approximate band-pass filters for economic time series (i.e.
how to calculate the ideal weights and the adjustment term) see Baxter and King (1995).
55
There is a trade-off when choosing the value of k: increasing k leads to a better approximation to the ideal filter, but results in more
lost observations. Baxter and King (1995) proposed a value of 3 or 6 to filter annual data.
from Penn World table compiled by Summers and Heston (1993), from 1960 to 1989, indicate that an
increase in inflation reduces the growth of GDP, other things being equal. Theoretical models which
founded a negative relationship between inflation and economic growth includes Jones and Manuelli
(1995), Wu and Zhang (1998) and Fernández Valdovinos (1999). Therefore, we expect that, in the long
run, an increase in the inflation rate will reduce output growth.
2. Government consumption: several papers have studied the empirical regularities relating fiscal
policies variables and the rate of growth of the economy. Some of those studies, like Engen and Skinner
(1992), found a consistently negative impact of the share of government spending on output growth rates,
lending support to the notion that smaller government sectors are associated with faster growth rates.
However, as pointed out by Aschauer (1989), when considering the impact of government spending on
output growth, it is important to distinguish between government capital accumulation and government
consumption. While the former one could have a positive impact on productivity growth, the later one
could entail distortions on private decisions leading to a lower growth rate. Thus, when considering the
long run, an increase (growth) in the ratio of government consumption to GDP will have a negative
relationship with output growth.
3. Investment ratio: in the neoclassical growth models of Solow (1956) and Swan (1956) an exogenously
higher value of the ratio of real gross investment to real GDP raises the steady state level of output per
effective worker and accordingly the growth rate tends to increase. For example, De Gregorio (1992),
using a five-year panel data for 12 Latin American countries between 1950 and 1985, finds that one of the
most important factors inhibiting growth in these countries was the low investment rate registered during
the period. Additionally, Bradford de Long and Summers (1991) found that machinery and equipment
investment has an strong association with growth and that this correlation is a much stronger one than
those found between growth and any other component of investment. Hence, an increase in the investment
in machinery to GDP ratio will result in the long run in a higher rate of output growth.
4. Private consumption: in the neoclassical growth models, a higher value of saving rate raises the
steady state level of output per capita and thereby increases the growth rate for a given starting value of
GDP. Thus, even though the saving rate does not affect long-run growth, for a given level of initial
income economies with higher savings rate will grow faster in the transition period. Accordingly, given a
level of income, a higher amount of private consumption means a lower saving rate and, therefore, a
lower growth rate.
5. Terms of trade shock: as stated by De Gregorio and Lee (1999), the terms of trade shock could be
considered as an exogenous variable that affects the growth rate of an individual economy. An
improvement in the terms of trade makes a country produce more and expand its export sector. Based on
data from Latin America countries, the regression results from these authors provide a significant positive
relationship between change in the terms of trade and per capita GDP growth. Thus, when considering the
long run, an increase (growth) in the terms of trade will have a positive influence on output growth.
6. Exports: over recent decades a considerable amount of empirical evidence has support the notion that
less protectionist regimes grows faster. In the early 1970´s Balassa and others began exploring the links
between trade and growth56. For example, Frankel and Romer (1999) use instrumental variables estimates
to analyze the effect of trade on income. Their results suggest that ordinary least squares estimates
understate the effects of trade, and that trade has a quantitatively large, significant, and robust positive
effect on income. Complementary, a large number of studies found that export growth and export levels
were highly correlated with GDP growth. See for example Edwards (1994) for a survey on this later
literature. Hence, in the long run we expect to find a positive correlation between the growth rate of
exports and GDP growth.
The data used in this section are from the International Monetary Fund, "International Financial Statistics"
and “Direction of Trade Statistics”, and from the Central Bank of Paraguay, “Boletín de Cuentas
Nacionales”. For every variable the original annual data runs from 1970 to 2000, so given the value
chosen for k, we have 19 observations for the filtered data. In Appendix A.2, Figures A.1.B. to A.8.B. plot
the long run relationship between the growth rate of GDP and 8 different variables. We have used the
filtered data and, for comparison, for each country we have also plotted the raw (original) data for the
period 1976-94 giving also a total of 19 observations for every variable (see Figures A.1.A. to A.8.A.).
For all the variables considered, the plots of the original data illustrate the absence of a clear relationship
over time between these variables and GDP growth. However, after filtering the data and extracting only
its long run components, a clear relationship between the two time series emerges. Complementary, Table
15 gives the correlations coefficients of the growth rate of GDP per capita and the different variables. This
table confirms the impression from Figures A.1.B. to A.8.B. The signs of the correlation coefficients are
the expected ones. For example, in the long run, the rate of GDP growth is negatively correlated with the
inflation rate, the government consumption to GDP ratio and the private consumption to GDP ratio. On
the contrary, a higher growth rate of GDP is observed with a higher investment in machinery to GDP
ratio, with a higher rate of growth of exports and with a favorable shock in the terms of trade.
56
See for example, Balassa (1978).
TABLE 15: Correlation Coefficients with GDP Growth Rate
Variables
Original Data
Filtered Data
Inflation rate
- 0.0997
- 0.8382
Gov. Cons./GDP
- 0.6038
- 0.5573
∆ (Gov. Cons./GDP)
0.0169
- 0.6716
Private Cons./GDP
- 0.5482
- 0.7506
∆ (Private Cons./GDP)
- 0.3424
- 0.9157
Investment/GDP
0.6904
0.8776
∆ Terms of Trade
0.1803
0.5424
∆ Exports
0.5389
0.8746
Source: Authors´ estimates using the Baxter and King filter.
6. The Co-movement of Paraguay with its trade partners.
In this part of the paper we address the issue of the cyclical co-movements of the Paraguayan output with
those from countries in the MERCOSUR area and the USA57. The main features of the aggregate
fluctuations in the seven countries are considered, exploring the direction and magnitude of the comovements of output across countries58. Furthermore, we study the association of their business cycles
decomposing the series in output into cycles of different frequencies. Due to its widespread use in
empirical economics, the Hodrick-Prescott filter is employed to mechanically decompose the individual
series into a trend movement and a cyclical component. Correlation analysis is then used to summarize the
extent to which the cyclical components exhibit co-movements across countries. Finally, developments
over time in the synchronization of the series cyclical component are examined on the basis of the
contemporaneous cross correlation coefficients for rolling 10-year periods.
6.1. Degree of economic integration.
Before analyzing the correlation of business cycles across countries, we first present some data on the
degree of integration between the different economies and Paraguay. We consider two widely used
indicators: the share in trade and the share in foreign direct investment. Paraguay´s major regional trading
partners are Mercosur, the European Union and the East Asian countries. For example, in 2001 the share
of Paraguay´s trade with Mercosur (including Chile and Bolivia) amounted to 59.2% of total trade, and
57
Throughout this section we shall refer as MERCOSUR to Argentina, Bolivia, Brazil, Chile, Paraguay and Uruguay.
the corresponding figures with the European Union and East Asian countries were 9.8% and 16.3%,
respectively. Table 16 below shows the degree of integration with the different countries in the Mercosur
area and with the USA. It can be seen that trade integration with the countries considered in this part of
the paper is quite high, specially with Brazil and Argentina. Brazil has an average share of 30.1% in total
trade (36.2% in exports and 28.0% in imports), while for Argentina this share reach 16.5% (7.9% in
exports and 18% in imports). The importance of the other countries in total trade is not very significant:
Bolivia 0.5%, Chile 2.4%, Uruguay 3.9% and USA 9%. Regarding the stock of foreign direct investment
in Paraguay, Argentina and Brazil account again for a large percentage of the total: 16.4% and 10.3%,
respectively. However, the share with USA is the largest, 34.2%. The contribution of the remaining
countries is negligible. Hence, according to the indicators presented, Paraguay´s integration with the
countries considered in this section ranges from small to considerably large.
TABLE 16: Share in trade and foreign direct investment. Average 1995-2001
Argentina
Bolivia
Brazil
Chile
Uruguay
USA
Total Trade
16.5%
0.5%
30.1%
2.4%
3.9%
9.0%
Exports
7.9%
1.3%
36.2%
3.9%
6.6%
4.7%
Imports
18.0%
0.1%
28.0%
1.8%
2.7%
10.5%
16.4%
n.a.
10.3%
1.16%
5.9%
34.2%
Foreign Direct Investment
Source: Author’s estimates.
6.2. Growth correlations across countries.
A first glance in the grade of symmetry between Paraguay and the other economies in the MERCOSUR
area and the USA can be obtained by analyzing unprocessed data from the countries. Annual data on real
GDP for the countries considered in the study, spanning the period 1970-2000, were obtained from the
International Monetary Fund (IMF) and the Economic Commission for Latin America and the Caribbean
(ECLAC). For each country, the rate of growth of GDP is calculated as the first difference of the
logarithm of real GDP.
We first consider data on the growth rate of real GDP. Table 17 shows standard deviations and correlation
coefficients for the growth in output across the seven countries, considering the full data period and also
ten years intervals. Correlations are measured with respect to Paraguay.
58
In Mundell´s theory of optimum currency areas the incidence of disturbances across regions or countries is a critical determinant in
the design of those areas. Fernández Valdovinos (2000) explores the feasibility of a currency area in MERCOSUR analyzing the
distribution of output disturbances across countries in the region.
TABLE 17: Output Growth Correlations and Volatilities.
Periods
Argentina
Bolivia
Brazil
Chile
Uruguay
Paraguay
USA
Correlations with Paraguay
1970-00
-0.1036
0.3761
0.4751
0.1889
0.3122
1.0000
-0.0317
1970-79
0.0587
-0.8366
-0.3055
0.3957
0.3408
1.0000
0.2783
1980-89
-0.4568
0.7035
0.2195
0.6807
0.5049
1.0000
-0.1246
1990-00
0.0833
0.4200
0.6539
0.4618
0.0460
1.0000
-0.5169
Standard Deviation
1970-00
0.0509
0.0310
0.0433
0.0616
0.0414
0.0389
0.0219
1970-79
0.0442
0.0176
0.0371
0.0707
0.0272
0.0229
0.0260
1980-89
0.0487
0.0279
0.0459
0.0705
0.0572
0.0444
0.0255
1990-00
0.0508
0.0162
0.0215
0.0359
0.0368
0.0170
0.0156
Source: Authors´ estimates.
For the full period, the data shows that the degree of output growth volatility have been very different, not
only across countries but also over time. For the whole period, USA is by far the country with the lowest
standard deviation (a value of only 0.022). In the MERCOSUR area, Bolivia and Paraguay present the
lowest degree of volatility (with 0.031 and 0.039 respectively), all other values for this statistic are above
0.04. Moreover, the data indicates that output fluctuations in MERCOSUR countries have generally been
bigger (across all countries) during the eighties than during any other ten years period. The findings also
show that for most countries, and certainly for the average, volatility of the growth rate of GDP reached
its lowest value during the nineties59. For Paraguay, the growth rates of real GDP have been relatively
stable in all periods with a coefficient of standard deviation above 0.04 only during the eighties.
On the other hand, correlation coefficients reveal that, when considering the whole period, Paraguayan
output growth was more highly correlated with the Brazilian one60. Additionally, for the whole period,
correlations with Bolivia and Uruguay reach a value of 0.38 and 0.31, respectively. In general, the
correlation coefficients are not particularly very high with values below 0.5, revealing a moderate degree
of co-movement of the different economies with Paraguay. However, given the observed instability of the
coefficients over time, it seems more plausible to examine those coefficients by breaking the sample in ten
years periods. The analysis by sub-samples have the correlation coefficients reflecting, during the
seventies, Paraguay’s dependence on USA, one of its main trading partner in this period. The construction
of the highway to Brazil, the development of the lands in the frontier region with this country, the
59
60
For Argentina and Uruguay, the seventies were the more stable years.
This result is explained mainly by the close behavior of the economies during the nineties.
building of the world’s largest hydroelectric project (Itaipú) and the implementation of MERCOSUR
substantially change the degree of co-movements with the countries. For example, the ensuing dramatic
increase in trade with Brazil was echoed in a higher value of the correlation coefficient during the eighties
and nineties61. Notice also the low degree of correlation of the Paraguayan economy with the USA.
During the eighties and nineties the correlation coefficient is even negative. On the other hand, output
growth correlations have being also relatively high with Bolivia and Chile, specially during the last two
decades considered.
We have also calculated an alternative measure for asymmetric output disturbances by estimating the
parameter γi,j defined as the standard deviation of the difference in the growth rate of GDP between
countries i and j, SD(∆yi - ∆yj). Thus, for countries where business cycles are symmetric and national
output move together, the value of this measure will be small. Table 18 presents the parameter γi,j
estimated using the full period and intervals of ten years, for both MERCOSUR countries and the USA.
TABLE 18: Parameter Gamma, j = Paraguay
Periods
Argentina
Bolivia
Brazil
Chile
Uruguay
Paraguay
USA
1970-00
0.0672
0.0396
0.0423
0.0663
0.0472
0.0000
0.0453
1970-79
0.0486
0.0389
0.0492
0.0651
0.0290
0.0000
0.0295
1980-89
0.0795
0.0317
0.0564
0.0518
0.0518
0.0000
0.0539
1990-00
0.0522
0.0179
0.0165
0.0319
0.0399
0.0000
0.0284
Source: Authors´ estimates.
For the full period, data reveals that MERCOSUR countries or the USA do not usually have a close
business cycle conformance with Paraguay. The value of the parameter γi,j is in general, for any of the
countries, higher than 0.04 with a maximum value of 0.067 for Argentina62. However, when examining
the behavior of the parameter gamma in ten years intervals, business cycles with some countries seems to
be more synchronized. As found before, dissimilarities of business cycles have been significantly lower
during the nineties than during any other ten-year period, especially with Bolivia and Brazil. The value of
the parameter γi,j in the decade was only 0.018 and 0.017, respectively. These values are very close to
those observed in some European Union (EU) countries (see previous footnote).
61
The outcome that closer international trade links result in more closely correlated business cycle across countries is found in
Frankel and Rose (1997) using data in 20 industrialized countries over 30 years.
6.3. Business cycles fluctuations across countries.
In this section of the paper we employ the methodologies of current business cycle research, so as to
explore the direction and magnitudes of the co-movements of the economies. For a particular economic
variable, long-term developments are reflected in the trend of the variable while cyclical movements are
determined as short-term deviations from this trend. Nevertheless, it can prove difficult in practice to
distinguish between trend and cycle. Consequently, studies of the business cycles still face the basic
problem of how to isolate those features in the data that are associated with long-term growth and those
related with business cycles.
As a result, to decompose each of the time series in output into a trend component and a cyclical
component, we employ the well-known Hodrick-Prescott (HP) filter. The HP filter is applied to the
logarithm of the series and the smoothing parameter λ is set equal to 100, a number commonly used for
annual data.
TABLE 19: Business Cycles Co-movements and Volatilities.
Periods
Argentina
Bolivia
Brazil
Chile
Uruguay
Paraguay
USA
Correlations with Paraguay
1970-00
0.0292
0.6101
0.1725
0.7166
1.0000
0.6907
-0.1259
1970-79
0.1349
0.3693
-0.1142
0.5190
1.0000
0.8006
0.4861
1980-89
-0.0105
0.9773
0.2608
0.8884
1.0000
0.7367
-0.2907
1990-00
0.2640
0.6443
0.7302
0.7818
1.0000
0.4793
-0.5720
Standard Deviation
1970-00
0.0439
0.0348
0.0363
0.0617
0.0391
0.0473
0.0209
1970-79
0.0337
0.0344
0.0384
0.0719
0.0287
0.0341
0.0235
1980-89
0.0413
0.0443
0.0471
0.0764
0.0600
0.0719
0.0255
1990-00
0.0555
0.0151
0.0234
0.0369
0.0182
0.0316
0.0142
Source: Authors´ estimates.
Table 19 presents the results. Initially we consider the volatility of the output cyclical
component. The calculations show that volatility in MERCOSUR countries has been higher than
in USA. When analyzing the full period, the lowest standard deviations in the sample of
62
In comparison, business cycles correlations have been higher in European Union (EU) countries. Fernández Valdovinos (2000)
found an average value for γi,j of 0.017, 0.019 and 0.020 during each of the three periods considered (1970-79, 1980-89 and 1990-98).
The countries analyzed are Belgium, Denmark, France, Germany, Italy and Netherlands.
countries considered are for Bolivia, Brazil and USA (0.035, 0.036 and 0.021 respectively). At
the same time, the coefficient is slightly larger for Paraguay, 0.039. Even when considering ten
years intervals almost the same conclusions are reached. In addition, it is usually the case that
volatility has been much higher in the countries during the eighties than in any other period, the
exception being Argentina. It is perceived also that, excluding the Argentinean economy, the
nineties were the more stable years.
Regarding the pattern of correlation among the series, statistics reveal that, for Paraguay, co-movements
of outputs with other MERCOSUR countries or the USA were not usually very high. Specifically, when
considering the full period, the highest values are the coefficients with Bolivia, Chile and Uruguay, 0.61,
0.72 and 0.69 respectively63. However, data analyzed by shorter periods reveals that co-movements of the
series could be stronger when considering ten-years periods. For example, during the eighties the
correlation coefficient with Bolivia is above 0.97 and, with Chile, it is 0.89 and 0.78 during the eighties
and the nineties respectively. With the USA, the coefficient value has a significant value only during the
seventies, 0.48, and it is even negative during the eighties and nineties64. Notice also that business cycles
correlations with the MERCOSUR countries are, in average, much higher during the nineties.
Figure 7: Ten-years Rolling Correlation Coefficients
with Paraguay
1.20000
1.00000
0.80000
0.60000
0.40000
0.20000
0.00000
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
-0.20000
-0.40000
-0.60000
-0.80000
-1.00000
ARGENTINA
63
BRAZIL
URUGUAY
Fernández Valdovinos (2000) obtains that, for the same European countries cited before, output correlations are generally much
higher. In addition, it is found that the degree of co-movement is more pronounced in subgroups of countries: Belgium, France, Italy
and Netherlands. For these countries the correlation coefficient is in average above 0.70.
64
These results cast doubts about the convenience of dollarizing the Paraguayan economy.
A question that may arise is if there have been changes in the output co-movements over time. One
possible explanation for the low correlations found previously is that they reflect the low co-movement
from earlier periods. Figures 7 and 8 illustrate the correlation coefficients between cyclical components
compiled for rolling 10-year periods. Similar findings are obtained for rolling periods of shorter length.
The figures display the rolling correlations of Paraguay with the rest of the countries.
It can be seen that co-movements of output among the three full members of MERCOSUR and Paraguay
follow similar patterns. In most cases, correlations tend to slightly fall from the initial periods of the
sample up to the beginning of the eighties. Since then they all abruptly decline and, in some cases, they
become even negative at the beginning of the decade. Finally, coincidently with the beginning of the
MERCOSUR area, the coefficients resume to increase until the end of the nineties. The degree of
synchronization of output cyclical movements among Paraguay and these countries usually achieves its
maximum at the end of the period considered. For example, during this decade the correlation coefficient
with Argentina reach a value of 0.46 while the same coefficient with Brazil attains a value of 0.82.
The behavior of the coefficients with the other countries do not follow a common pattern. For the whole
period, correlation coefficients with Bolivia and Chile fluctuate between 0.40 and almost 1. Notice
however that the correlation coefficient with USA shows a clear tendency to decline over time and it has a
negative value since the mid eighties.
Findings in this section reinforce previously reached conclusions. Business cycles correlations with the
countries in the MERCOSUR area and the USA are not very high when considering the whole sample.
However, when disentangling the sample in ten-years periods sub-samples, more common patterns
emerge, especially with Bolivia, Brazil and Chile.
Figure 8: Ten-years Rolling Correlation Coefficients
with Paraguay
1.20000
1.00000
0.80000
0.60000
0.40000
0.20000
0.00000
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
-0.20000
-0.40000
-0.60000
-0.80000
BOLIVIA
CHILE
USA
1995
1996
1997
1998
1999
2000
7. Conclusions
In this work we have investigated the process of economic growth in Paraguay from the 1940s to the
present. The paper explored a variety of dimensions that are relevant to understand the recent economic
history of the country. The picture that has emerged is not an optimistic one and, if the trends that we
found are to be perpetuated, the future for Paraguay is grim and the country would remain as one of the
poorest countries in the hemisphere.
The three main conclusions from the case study in Paraguay are the following. First, there have been a
declining (or at best stagnant) path for the total factor productivity of the economy. Thus, despite a
significant accumulation of physical capital, income per capita has been unable to grow. Second, it could
be seen a poor performance of Paraguay in the accumulation of human capital. Compared with the region,
Paraguay has been left behind as vividly shown by the statistics on poverty. Third, the paper confirmed
the importance of macroeconomic stability and of aggregate fluctuations with the country’s main trading
partners. Our results indicate that periods of macroeconomic stability have been associated with higher
growth. Furthermore, the data shows an increase over time in the importance of fluctuations with some of
the country’s trading partners, specially Brazil.
We believe that the first two findings, lack of productivity growth and human capital accumulation, are
the results of a highly inoperative public sector. It is important to highlight this point as the (relatively)
small size of its public sector is perhaps the most distinguishing feature of the Paraguayan economy in the
Latin American context. As any relevant theory of economic growth would predict, small taxes, specially
on capital, are likely to foster growth. However, what appears to be the dominant effect, investments in
infrastructure, could radically determine the rate of return of private investment. We believe that the
government in Paraguay has been subpar in the provision of public investment.
A similar consideration can be made on the accumulation of human capital. There are numerous reasons
to believe that unregulated markets would under perform in the provision of quantity and quality of
primary schooling, which is simply the first step in the accumulation of human capital and productive
skills. Thus, in the presence of a public sector inefficiently providing primary and secondary education, it
can only be expected that the country would not take off. It is also important to emphasize that, as human
capital is complementary to physical capital, the lack of its accumulation would also impaired the
accumulation of physical capital.
Improving the accumulation of infrastructure and human capital would require in the case of Paraguay of
a pivotal role by the government in the future. To that end, the collection of taxes need to be improved.
But even then, the Paraguayan may not be able to succeed by itself and the international community may
have an important role to play.
In terms of the relevance of macroeconomic stability, the lessons are in line with the current consensus
and, thus, we do not need to elaborate on them further. In terms of the higher interdependence on the
Brazilian economy, we just want to emphasize that this may simply be the outcome of geographical,
historical and cultural proximity, which are natural advantages that Paraguay should definitely exploit.
However, it seems safe to assert that in addition to these static gains from trade advantages, Paraguay
would benefit greatly in terms of growth and stability if it reduces any existent biases in the trade and
investment with the rest of the world.
Appendixes
A.1. Incentives to accumulate capital.
Paraguay´s tax/GDP ratio has been usually among the lowest in the hemisphere, remaining usually below
10% since 1971. Although this implies that the tax burden of the private sector was low, the limited
volume of public sector resources had also some drawbacks: public sector wages were usually low (which
encourage corruption); public investment in some of the standard public sector areas, such as transport,
basic health and education, has been limited and social expenditures benefiting the poor were low.
TABLE A.1: Paraguay - Tax Burden. In % of GDP.
1970
1973
1975
1978
1980
10.3
8.0
8.1
9.4
8.4
1983
1985
1988
1990
1993
6.4
6.9
6.9
9.0
8.5
1995
1998
2000
10.0
10.6
9.9
Source: Central Bank of Paraguay.
In 1992 a significant, and at the time largely overdue, tax reform was implemented in the country. The tax
system in force until 1991 was characterized by the proliferation of legal norms and an immense quantity
of taxes and tax rates, which indeed constituted a set of tax laws which were very complex and difficult to
manage65. Among its most salient features, we may mention the following characteristics:
•
Predominance of indirect taxes, which for the most part responded to fiscal needs of a partial
nature.
•
The regressive nature and complexity of the system, which encouraged fiscal evasion and
increased the loss of credibility towards the administrative taxation agency.
•
The proliferation of tax exonerations and special regimes.
•
Custom tariffs that were contrary to a policy of openness and integration.
•
Administrative bureaucracy which provided incentives for the infringement of tax laws by
evaders.
65
According to multilateral organizations, Paraguay’s tax system at that time was antiquated and very inefficient. It was
often argue by these institutions that Paraguay’s tax system has not kept pace with domestic inflation and growth, a fact
which endangered macroeconomic equilibrium, future growth and prospects for eradicating poverty.
Before 1992, taxes could be classified into four broad categories: taxes on goods and services, income
taxes, taxes on capital and foreign trade taxes. The first category grouped sales taxes, several selective
consumption taxes (fuel, liquor, cigarettes, livestock, etc.), stamp taxes on different kinds of transactions
and several other small taxes. In term of revenues, they represented about 4.2 – 4.3 percent of GDP in
1984-88.
The structure of taxes on goods and services changed significantly overtime. The general sales tax
represented 0.6 percent of GDP in 1984, a ratio that increased progressively to 0.8 percent by 1987.
Although exemptions were widespread and potential revenues were difficult to calculated, evasion must
have been large, since the tax rate on domestic sales was 4 percent, and on imports 8 percent or 14
percent, with 80 percent of the proceeds coming from the latter two66. The domestic sales tax was charged
only to the final consumer. Even if it affected only half of GDP, it should have generated more than 2
percent of GDP in revenues but collections were less than half of that. Even though selective consumption
taxes and stamp taxes were classified under this heading, they include many taxes with no relation to each
other. For example, the stamp tax included 84 different taxes affecting civil and commercial dealings.
Many of these taxes were specific and thus declined in importance with inflation. Erosion of potential tax
revenues also occurred through widespread exemptions.
TABLE A.2: Paraguay – Tax Structure. In % of GDP.
Taxes
1984
1986
1988
1990
Taxes on Goods & Services
4.21
4.18
4.30
4.61
2.26
2.40
2.30
2.21
General Taxes
0.56
0.74
0.81
0.78
Selective taxes
1.70
1.66
1.49
1.43
Stamp Taxes
1.79
1.65
1.90
2.34
Other
0.15
0.13
0.10
0.06
Income taxes
1.11
1.25
1.42
1.26
Capital Taxes
0.38
0.38
0.28
0.27
Land / Property
0.35
0.35
0.25
0.27
Other
0.03
0.03
0.03
0.00
Other taxes
1.01
1.01
1.02
3.08
Tax Revenues
6.71
6.82
7.02
9.21
Consumption
1
1
Mostly import and export taxes.
Source: Central Bank of Paraguay.
66
Therefore the sales tax basically amounted to an import tax.
The second broad category of taxes, income taxes, was another collection of uncoordinated small taxes.
They applied mainly to enterprises, since the personal income tax applied in few instances and was
negligible in effect. The income tax on profits (agriculture was exempted) was slightly progressive, with
rates moving from 25 percent to 30 percent with higher corporate incomes. Incomes taxes represented 1.6
to 1.7 percent of GDP in the early 1980s, with that share dropping to 1.2 percent in 1984-86; they were
about as much in 1990. Evasion must have been pervasive in this category as well. With returns to capital
amounting to about half of value added and assuming that the tax applied to half of GDP, the enterprise
tax alone should have represented 6.5 percent of GDP, over 5 times the actual collection. It must be
mentioned that an Investment Incentives Law was enacted in 1990. This law gave beneficiaries 5-year tax
holidays on 95 percent of income taxes and 6 months of duty-free imports.
Additionally, capital taxes generate little revenue, about 3 percent of total taxes and less than 0.3 percent
of GDP. Although tax rates were about 1 percent of property value, assessments were extremely low. On
the average, urban property tax values represented less than 35 percent of market value and in rural areas
taxable values were just 5 percent of market values. The capital tax category also included an inheritance
tax that was so easily evade that its proceeds were insignificant.
Regarding trade taxes, during these years Paraguay operated with tariff that were low and quite
homogeneous, despite a Custom Law often suggesting high tariff (30 percent and sometimes reaching
over 70 percent) with wide dispersion. Three factors had helped achieved low, homogenous tariffs in
practice. First, simple special regimes operating with low flat rates have replaced many ordinary tariffs.
Second, taxes often reaching close to 5 percent of imports were charged under different names even on
tariff-free items; therefore, even though the lowest tariff was often zero, some taxes were paid even in
these cases. And third, unregistered imports set a ceiling on tariff rates. If tariffs exceeded 10-15 percent,
goods tended to be imported through informal channels.
As mentioned before, taxes were widely evaded in Paraguay during this period because of what the
private sector considered unreasonably high tax rates: 30 percent tax on profits, some high import tariffs,
and the stamp tax (an inefficient scheme that levies contracts rather than output, income or wealth). To
avoid these taxes, the private sector developed a complex “parallel economy” with surprisingly favorable
results relative to those obtained in the “formal economies” in neighboring countries. Additionally,
sanctions for failing to pay taxes due on time varied by tax and they often were nonexistent. In fact, the
system not only did not penalize infractions, but in practice encouraged tax payers to avoid payments. In
most instances, penalty interest rates were lower than commercial interest rates. Thus, for the taxpayer, it
was more profitable to delay payment until the infraction was discovered (if it ever was) than to pay on
time.
In 1992, the authorities proposed a plan to reduce the number of taxes while making the system simpler
and easier to manage. It basically: (i) replaced the stamp tax and a myriad of small, difficult to collect,
indirect taxes with two new ones: a value added tax that replaced sales taxes and a few ad-valorem taxes
on consumption (fuels, liquors, cigarettes, luxuries, etc.); (ii) enacted a direct and indirect “sole” tax on
small businesses replacing all direct and indirect taxes applying to them67; (iii) increased the profits tax
from a progressive system involving two rates with a flat rate equal to the higher prior rate; (iv) enacted
an income tax on agriculture property using the presumptive income concept; and (v) changed the penalty
system as needed to truly fight late payments and evasion. Other reforms included proper assessment of
property values and the abolition of the inheritance tax.
TABLE A.3: Paraguay – Tax Structure. In % of GDP.
Taxes
1995
1997
1999
2001
Income taxes
2.3
2.1
2.3
1.5
Taxes on Goods & Services
6.1
6.3
6.0
6.4
1.2
1.3
1.3
1.9
Value added tax
4.4
4.5
4.3
4.1
Stamp tax
0.4
0.4
0.3
0.3
Other
0.2
0.1
0.1
0.2
2.8
2.3
1.7
1.7
2.8
2.3
1.7
1.7
11.2
10.7
9.9
9.6
Excises taxes
Taxes on international transactions
Import duties
Tax Revenue
Source: Central Bank of Paraguay.
The tariff regime improved substantially after the reform and Central Government revenues increased
rapidly during the nineties, reaching around 10% of GDP. Most of the increase in revenues was due to
import taxes and the introduction of a value-added tax of 10 per cent. Import taxes, which today account
for around 20% of tax revenues, increased from 0.9% of GDP to 2.8% of GDP in 1995, slightly
decreasing afterward to 1.7% of GDP in 2001. At the same time, the value added tax accounted for an
equivalent of 4.4% of GDP in 1995 and 4.1% of GDP in 2001. Another important source of revenue came
from the income taxes and from excises taxes in some specific products. Somewhat surprisingly,
Paraguay was able to collect revenues worth 10% of GDP with a comparatively low tax burden and few
67
Small enterprises were exempt from the value added tax, but they paid the “sole” tax mentioned above. Additionally,
they allowed to deduct from that “sole” tax half the amount of the value added tax they paid on the inputs they
purchased.
forms of taxation: a VAT of 10 per cent, low import tariffs, no personal income tax and a corporate
income tax of 30 per cent with many exceptions.
It must be mentioned that, under the Treaty of Asunción of March 1991 that was ratified in July 1991,
Paraguay agreed to an automatic schedule of tariff reduction as well as to reduce its list of exceptions so
as to become part of the MERCOSUR free trade zone. Additionally, beginning in July 1992 Paraguay
replaced the tariff schedule enacted in July 1991 with a new schedule of even lower and more
homogeneous rates and instituted a value-added tax on imports at a 10 percent uniform rate. The tariff
changes left tariff positions –not including lists of exceptions and internal consumption taxes—at 0
percent for inputs, 5 percent for capital goods, 10 percent for final goods and 15 percent and 20 percent
for cars68. At the same time, imports were also subject to a 10 percent value-added tax. These changes
brought the tariff code more in line with the de facto openness of the economy.
In summary, during the whole period, tax and external tariff distortions to accumulate capital seem to
have been not important in Paraguay. For the years before 1992, tax evasion, special tax and tariff regimes
and smuggling substantially reduced the tax burden of the private sector. After the tax reform, even
though initially it was expected to yield higher revenues, the tax/GDP ratio remained among the lowest in
the region. It is well known that incentives for investments, such as exemption from taxes, subsidies and
other benefits, can be important, but may not be crucial in attracting private capital. Paraguay seems to be
an example. The Paraguayan Investment Incentives Law provides better incentives than the laws of many
Latin American countries and it still could not attract large flows of private capital.
As agreed by several authors, an important element to attract private capital seems to be the maintenance
of stable policies in the long run, which should be enforced by the credibility of the country. As
mentioned by Insfrán Pelozo (2001), this credibility could be obtained in the long run only by restraining
the government´s ability to change the rules of the game with respect to restrictions on capital movements,
taxes, property rights, risk of expropriation, non convertibility of local currency, civil wars, etc. Similarly,
Barro (1991) and De Gregorio and Lee (1999) mention the rule of law and the quality of political
institutions as important factors explaining growth rates across countries. They consider that institution
environment that secures property rights and provides a strong legal system is central for investment and
other aspects of economic activities. We may need to look at all these alternative factors in order to
understand why growth and investment have been so low in Paraguay.
68
The tariff simplification eliminated all special regimes and exceptions.
A. 2. Graphs for Growth and Other Main Macro Variables.
Figure A.1.B. Inflation and Growth. Filtered data.
GROWTH
GROWTH
Figure A.1.A. Inflation and growth. Unfiltered data.
INFLATION
INFLATION
Figure A.2.B. Gov. Cons./GDP. Filtered data.
GROWTH
GROWTH
Figure A.2.A. Gov. Cons./GDP. Unfiltered data.
GC/GDP
GC/GDP
Figure A.3.B. ∆ (Gov. Cons./GDP). Filtered data.
GROWTH
GROWTH
Figure A.3.A. ∆ (Gov. Cons./GDP). Unfiltered data.
CHANGE IN (GC/GDP)
CHANGE IN (GC/GDP)
Figure A.4.B. Private Cons./GDP. Filtered data.
GROWTH
GROWTH
Figure A.4.A. Private Cons./GDP. Unfiltered data.
PC/GDP
PC/GDP
Figure A.5.A. ∆(Private Cons./GDP). Unfiltered data.
GROWTH
GROWTH
Figure A. 5.B. ∆(Private Cons./GDP). Filtered data.
CHANGE IN (PC/GDP)
CHANGE IN (PC/GDP)
Figure A. 6.B. Investment/GDP. Filtered data.
GROWTH
GROWTH
Figure A.6.A. Investment/GDP. Unfiltered data.
INV/GDP
INV/GDP
Figure A.7.B. ∆ Terms of Trade. Filtered data.
GROWTH
GROWTH
Figure A.7.A. ∆ Terms of Trade. Unfiltered data.
CHANGE IN TERMS OF TRADE
CHANGE IN TERMS OF TRADE
Figure A.8.B. ∆ Exports. Filtered data.
GROWTH
GROWTH
Figure A.8.A. ∆ Exports. Unfiltered data.
CHANGE IN EXPORTS
CHANGE IN EXPORTS
References.
•
Artis, M. and Zhang, W. (1999). “Further evidence on the international business cycle and
the ERM: Is there a European business cycle?”. Oxford Economic Papers 51, pp. 120-132.
•
Aschauer, David (1989). “Is public expenditure productive?”. Journal of Monetary
Economics 23, pp. 177-200.
•
Baer, W. and Breuer L. (1986). “From Inward to Outward Oriented Growth: Paraguay in the
1980s”. Journal of Inter American Studies and World Affairs , Vol. 28, Issue 3, pp. 125-140.
•
Baer, W. and Birch M. (1984). “Expansion of the Economic Frontier: Paraguayan Growth in
the 1970s”. World Development, Vol.12, No.8, pp. 738-798.
•
Balasa, Bela (1978). “Exports and Economic growth: Further evidence”. Journal of
Development Economics, Vol. 5, No.2, pp.181-189.
•
Barro, Robert (1991). “Economic growth in a cross section of countries”. Quarterly Journal
of Economics, Vol. 106, No.2, pp.407-443.
•
Bayoumi, T. and B. Eichengreen (1997). “Optimum currency areas and exchange rate
volatility: Theory and evidence compared”, in B.J. Cohen (Ed.), International trade and finance new
frontiers for research: Essays in honor of Peter Kenen. Cambridge University Press, Cambridge, Mass.
•
Bayoumi, T. and B. Eichengreen (1993). “Shocking aspects of European monetary
unification”, in F. Torres and F. Giavazzi (Eds.), Adjustment and growth in the European Monetary
Union. Cambridge University Press, Cambridge, Mass.
•
Baxter, M. and R. King (1995). “Measuring business cycles approximate band-pass filters for
economic time series”. National Bureau of Economic Research Working Paper No. 5022.
•
Bradford de Long J. and L. Summers (1991). “Equipment investment and economic
growth”. The Quarterly Journal of Economics, Vol. CVI, pp. 445-502.
•
Burns, A. and W. Mitchell (1946). “Measuring business cycles”. National Bureau of
Economic Research. New York, New York.
•
Cabello, C., A. Hosono and J. Molinas (2000). “Escenarios alternativos de crecimiento de la
economía paraguaya”. Unpublished Manuscript.
•
De Gregorio, José (1992). “Economic growth in Latin America”. Journal of Development
Economics, Vol. 39, pp. 59-84.
•
De Gregorio, J. and J. W. Lee (1999). “Economic growth in Latin America: sources and
prospects”. Paper prepared for the Global Development Network Research Project.
•
Economic Commission for Latin America and the Caribbean (1981). “Statistical Yearbook
for Latin American and the Caribbean. Edition 1981”. New York, New York.
•
Economic Commission for Latin America and the Caribbean (2001). “Statistical Yearbook
for Latin American and the Caribbean. Edition 2001”. New York, New York.
•
Edwards, Sebastian (1994). “Openness, trade liberalization and growth in developing
countries”. Journal of Economic Literature, Vol. 31, No. 3, pp. 1358-1393.
•
Eichengreen, Barry (1998). “Does MERCOSUR need a single currency?”. National Bureau
of Economic Research Working Paper No. 6281.
•
Engen E. and J. Skinner (1992). “Fiscal policy and economic growth”. National Bureau of
Economic Research Working Paper No. 4223.
•
Fernández Valdovinos, Carlos (2000). “Cyclical co-movements in output across
MERCOSUR countries”. Working Paper. Central Bank of Paraguay.
•
Fernández Valdovinos, Carlos (1999). “Inflation and welfare in an endogenously growing
economy”. Unpublished dissertation. University of Chicago, Chicago, Illinois.
•
Fischer, Stanley (1993). “The role of macroeconomic factors in growth”. National Bureau of
Economic Research Working Paper No. 4565.
•
Frankel, J. and D. Romer (1999). “Does trade cause growth?”. The American Economic
Review, Vol. 89, pp. 379-399.
•
Frankel, J. and A. Rose (1997). “Is EMU more justifiable ex post than ex ante?”. European
Economic Review 41, pp. 753-760.
•
Hall, L.;
A. Hoffmaister; A. Monge and E. Robles (2000). “Fluctuations, Trends and
Transmission of Shocks in Central America, Mexico and U.S.A.”. Unpublished working paper.
•
Insfrán Pelozo, José Aníbal (2001). “Privatization in Paraguay: Some lessons”. Quarterly
Review of Economics and Finance, Vol. 41, Nº 5, 2001. pp. 737-753.
•
International Monetary Fund (1997). “Paraguay: Country Report”. No. 01/87. Washington
D.C.
•
International Monetary Fund. Public “Information Notices”. Several Numbers. Washington
D.C.
•
Jones, L. and R. Manuelli (1995). “Growth and the effects of inflation”. Journal of Economic
Dynamics and Control, Vol. 19, pp. 1405-1428.
•
Lucas, Robert (1980). “Two illustrations of the quantity theory of money”. The American
Economic Review, Vol. 70, pp.1005-1014.
•
McKinnon, Robert (1963). “Optimum currency areas”. The American Economic Review 53,
pp. 717-724.
•
Mundell, Robert (1961). “A theory of optimum currency areas”. The American Economic
Review 51, pp. 657-665.
•
Psacharopoulos, George (1994). “Returns to investment in education: A global update”.
World Development, Vol. 22, No. 9, pp. 1325-1343.
•
Robles, Marcos (1999). “Pobreza y Distribución del Ingreso en Paraguay, 1997/98”. Informe
de Consultoría. Dirección General de Estadísticas y Censos, Secretaria Técnica de Planificación del
Desarrollo Económico y Social. Asuncion, Paraguay.
•
Rojas, Bernardo Darío (2001). “La Influencia de Itaipú en la economía paraguaya”.
Unpublished manuscript.
•
Solow, Robert (1956). “ A contribution to the theory of growth”. Quarterly Journal of
Economics, Vol. 70, No. 1, pp. 65-94.
•
Summers, R. and A. Heston (1993). “Penn World Tables, Version 5.5”. available on diskette
from the National Bureau of Economic Research.
•
Swan, Trevor (1956). “Economic growth and capital accumulation”. Economic Record, No.
32, pp. 334-361.
•
United Nations Development Program (2001). “Informe sobre Desarrollo Humano 2001”.
New York, New York.
•
United Nations Development Program (2003). “Informe Nacional sobre Desarrollo Humano,
Paraguay 2003”. Asuncion, Paraguay.
•
World Bank (2001). “Paraguay. Attacking poverty”. Unpublished manuscript. Washington
D.C.
•
World Bank (1999). “Paraguay: Country Economic Memorandum”. Macroeconomic Policies
to Reactivate Growth. Washington D.C.
•
World Bank (1993). “Paraguay: Country Economic Memorandum”. Washington D.C.
•
World Bank (1992). “Paraguay: Country Economic Memorandum”. Washington D.C.
•
Wu, Y. and J. Zhang (1998). “Endogenous growth and the welfare costs of inflation: a
reconsideration”. Journal of Economic Dynamics and Control, Vol. 22, pp. 465-482.
***********